hey guys please find your seats we're
going to get started it's great to see
you all we have a very exciting topic
today with some very exciting speakers
I'm so excited to be here with you to
talk about AI competition and startups
uh before I recognize our spe speakers
and special guests from Washington I
want to tell you a little bit about why
we're here and YC's public policy
program last October we hired Luther low
to be yy's first head of public policy
where is he
all right you'll you'll hear more from
him later why does YC have to have like
why should YC have a DC presence it's
pretty simple for the longest time only
the largest players in the tech
ecosystem have had a seat at the
table but to paraphrase paricles you may
not be paying attention to politics but
politics is paying attention to you YC
wants to fight for little Tech in
Washington and make sure all of you have
a seat at the table one of the most
interesting spaces where we we're seeing
a lot of fierce competition is at the
model layer of AI today we're going to
be speaking with the top Anti-Trust
enforcers whose job it is to keep the
markets free and fair joining us from
Washington our FTC chair Lena Khan and
doj antitrust head Jonathan caner these
two are probably best known as Joe
Biden's trustbusters and Lead antitrust
enforcement for the FTC and doj combined
they oversee literally thousands of
attorneys and nearly a billion dollars
per year of budget and of course we will
end things with a fireside chat with our
very own state senator Scott weiner we
also have some important Partners I want
to recognize for this event including
the Omar Network electronic Freedom
foundation and Mozilla who will be on a
panel later and finally it wouldn't be a
YC event if we weren't ultimately
bringing it back to our Founders so in
between uh the these presentations we'll
actually have amazing demos uh
exhibiting some of the incredible stuff
that you all are building as we stand at
the Forefront of the AI Revolution
understanding the regulatory landscape
is crucial for every founder in this
room and that's why I'm thrilled to
introduce our next speaker Jonathan
caner and as Assistant Attorney General
for the antitrust division at the US
Department of Justice is at the center
of shaping how AI will be regulated and
how competition in this space will
unfold he's currently overseeing major
antitrust cases that could significantly
impact the AI ecosystem his insights on
antitrust concerns in AI data monopolies
and F fostering Fair competition in this
rapidly evolving field will be
invaluable as you build and scale your
AI startups so without further Ado
please join me in welcoming Jonathan
caner to discuss the future of AI
regulation and
competition thank you Gary and thank you
to YC and and M arzil our hosts for
today um I'm thrilled to be here and so
um I have some prepared remarks and I'll
deliver them but I I think it's
important for me to kind of start really
from the heart which is I'm an antitrust
enforcer um and so what does that mean
what does that mean about the work that
you do um I want to be very clear about
this our goal is to make sure that you
can build businesses that succeed that
Thrive and that take on new ideas new
problems and solve them right anti trust
enforcement is about making sure that
there is opportunity for people to have
great ideas to build funding around
those ideas and to succeed that's our
goal our goal is to make sure that the
markets in the marketplace has room for
all the people who are building
businesses for all the people who are
starting up companies for all the people
who are daring to take on incumbents and
and build new products and services that
are disruptive that's successful and in
a nutshell that's what we're doing and
that's why we're here it's no secret
that AI is creating a transformative
moment it's wonderful uh technological
revolutions are inflection points that
lead to Major change and major
opportunity it is an invitation for for
new technology and two technological
innovations and platforms are invitation
for people to invest to build um to
create that is wonderful it is a um
exciting time and AI is one of those
really important Transformations but
it's important to make sure we get it
right and it's important to make sure
that uh these Technologies are not only
safe and sound but that there is enough
opportunity uh so that businesses of all
sizes in all places throughout the
country have a fair opportunity to
compete on the merits of their
Innovations and on their um skill um and
so um it got me
thinking um about regulation versus
enforcement and I wanted to clarify that
so I'm a law enforcer I'm not a
regulator my job isn't to regulate a
market it's to enforce the law um only
when companies violate it to make sure
that competition can Thrive I went back
and was thinking about uh justice Lewis
brandise someone that is near and dear
to my heart there is something he said
in 1912 that I think is very important
for
today um in an address he said he asked
a question same question I think that I
want to present to you all to think
about today shall we regulate
competition or shall we regulate
Monopoly it was an incisive and astute
question
then at the time there were a range of
political leaders that believed um that
Monopoly monopolies were itable and so
we should let them emerge and then we
should just regulate the heck out of
them and have very um strong oversight
but Justice brandise rejected that idea
he said and these are quotes private
Monopoly in industry is as he said never
desirable and it is not inevitable
monopolies were not inevitable then and
monopolies are not inevitable now the
premise of the question presented by
Justice brandise is that is it more
desirable to rely on
competition and Market forces than the
inevitability of concentrated markets
where the only remaining solution is to
oversee them with invasive regulation in
all markets but especially markets
involving technology and key inflection
points the first line of defense is
competition as Justice Brandy said no
syst system of Regulation can safely be
substituted for the operation of
individual liberty as expressed in
competition to do so he said quote
attempting to substitute a regulated
monarchy for a public um would be the
result and so today there's no
substitute for the competitive process
competition is what ensures continued
Innovation and that improvements are not
hampered
or impeded by a monopoly choke points
competition is what ensures that
consumers entrepreneurs workers benefit
in the long run it ensures the freedom
of opportunity that is available to all
including and especially companies that
invest in new business and new business
ideas strong competition and
preservation of competition especially
early
on um can go a very long way promoting
competition early on enable as many
Innovations as possible to survive and
flourish promoting competition early on
can sure that it's not only the big
companies that succeed but also the
small businesses and the startups that
have the opportunity to grow promoting
competition early on can avoid the more
drastic regulatory measures down the
line that would be necessarily necessary
once monopolies emerge and the
importance of competition in AI
at the early stages cannot be
overemphasized we must examine Market
realities and promote sound policies
that promote and sustain opportunities
for free and Fair competition for both
AI infrastructure AI applications
Foundation models promoting competition
now will go a long way because it will
shape how they develop and create oxygen
for everybody to breathe how companies
develop AI applications and how those
applications
are ultimately brought to Market will
shape the field the state of competition
in AI hardware and chips for example
will have enormous consequences for
competition throughout the AI
ecosystem how we strike the balance
between ai's use of creators content and
providing just compensation for those
creators will shape the future of all
industry if we can protect competition
AI holds the promise of generating new
businesses and new markets that can
truly form the new Main Street both
figuratively and
literally little tech companies small
medium-sized businesses have the
potential throughout the country to
become the new version of The Corner
Store being built on Main Street and
that Main Street doesn't have to just be
here in San Francisco it could be
anywhere throughout the country that is
the world that is the promise that we
are trying to preserve and startups that
build on new and emerging AI can create
small businesses and tools that support
small businesses ecosystems that don't
just work for the biggest companies with
massive global scale but work for other
corner stores and because it supports
business anywhere has potential to lead
to the dispersion of opportunity so that
we can see communities Thrive not just
urban communities but rural communities
and that will create new job
opportunities and lead to a rising tide
that lifts all boats so we must do
everything we can to promote competitive
markets especially as these Technologies
develop we have to ensure that the
markets are deconcentrated and that
opportunities are
dispersed We should strive for an even
playing
field um that allows for as many
companies and as many ideas as possible
to be tested created improved and paved
the way for new businesses and job
opportunities to achieve all of this the
first step is a deep understanding of
Market realities how does the stuff
actually work in May for example we held
a really exciting workshop at Stanford
uh along with Stanford University uh
promoting competition in AI where we
sought to hear from diverse perspectives
and we did from Scholars but also
content creators VCS
startups um in all different areas of
the economy everything from tech for
tech to Tech for healthcare those
conversations at least for me were
invaluable because it's how we learn
it's how we grow it's how we deepen our
understanding of these issues so that we
can have sound policy and make sound
policy choices we've also worked closely
with other agencies to develop our
thinking and understanding just this
past Tuesday we released a joint
statement on competition in generative
AI Foundation models and AI products
along with our counterparts at the FTC
which you hear from later today the EC
and the UK through this joint statement
we pledged our use of available powers
to promote effective competition in AI
so that all companies from the little
ones to the big ones and everyone in
between will have the opportunity to
succeed and we're going to continue to
make every effort and that means um
healthy markets that means uh room for
all business models whether they are
closed Source open source making sure
that business model especially open
source business models have the
opportunity to build platforms and stay
open and stay extensible um um so that
they can allow um the rising tide to
lift all boats and we can see the
promise of innovation be realized um
across the entire uh Continuum and help
form the new Main Street so with that
I'm really excited to sit down have a
conversation about some of these issues
with Gary and we'll go from there thank
you
[Applause]
Jonathan thanks so much thank you it's
great to be here I think we just I just
I just have only a few questions and I
think we're both hoping to like open it
up to you all but just to kick it off I
mean the FTC has received a lot of uh I
mean really attention and praise for
their statement on sort of importance of
protecting competitiveness in open
source especially at the foundation
level you know what does protecting
competitiveness sometimes mean basically
doing nothing or you know just wait to
see how the market plays out how do you
think about that so I think we have to
think about first of all we need to
follow the facts and the law wherever it
takes us and so if if there no problem
to address we don't need to step in and
try to address it for its own sake right
our goal is simply to say okay what's
happening in the marketplace and does it
necessitate action uh in the vast
majority of instances we don't do
anything whether it's um mergers or
conduct there's a lot of business that
happens without us having any anything
to do with it we focus on the instances
where there are bottlenecks to
competition uh where there are
impediments to the kind of development
and growth that we were talking about
early um that I was just talking about a
few moments ago uh in your opinion is
the AI Market at the foundation layer uh
competitive concentrated or something
else so we're we're seeing it play out
and so from my perspective um we have
the promise for competition uh and
that's really exciting and so you know
certainly we've seen the develop
velopment of um proprietary um
Foundation models and then we've seen as
I alluded to my remarks the development
of Open Source which has um a which can
be a force multiplier for competition
and that's really exciting um I think
one of the things that we talked about
our AI statement one of the things I've
talked about is making sure that open
store stays open right so that um people
continue to build on it and people can
continue to um um invest in the value
can um and the dependencies that are
created that value can be created and
realized by small and mediumsized tech
companies who are innovating on that to
me that is competition in all its forms
everything from proprietary to open um
and so a little bit of work early on by
us uh and our communities to make sure
that those models can Thrive to make
sure that they can succeed can go a very
long way to ensuring that there is not
just competition at the model level but
competition for those who develop on top
of models can you con comment on how AI
connects to you know the two US versus
Google lawsuits and your USV Apple
lawsuit that uh was filed in the spring
so I can't talk about our Active cases
but what I can say is that we want to
make sure that the next technological
inflection points particularly
surrounding AI have the opportunity to
usher in new generation of competitors
our cases and if you read them carefully
they're designed to protect the ability
of others especially um new Innovative
startups to compete and realize the
fruits of their Investments those are
the the underlying um um Concepts that
animate our cases and so um to the
extent that um many of the new
technological inflection points that are
emerging now are based on technologies
that revolve around AI um to me it's
important that we make sure that not
just we correct the Ys of the past but
that we create opportunity for those new
and emerging Technologies many of which
are being created by people in this room
have the opportunity not just to see the
light of day but to realize their full
competitive and economic potential with
that let's open it up for questions
thanks uh my name is Eric um really
appreciate the work that you've been
doing on the Apple front uh I started
two companies that are relevant in that
space a smartwatch company and a
messaging company I know that you can't
comment directly on the case but for
people who have uh let's just say much
less of a background what can we expect
like what are the major Milestones that
you see over the over the coming you
know months years in that uh in that
type of a case for example um and then
the second question is are there can you
think of any examples of enforcement
actions that you might ask the courts in
in this or or s or similar cases sure so
um let me start kind of with an
animating principle that cuts through a
lot of our cases and you can see it in
some of the cases you mentioned you can
see of some of our recent cases
involving concert tickets and other a
lot of our cases focus on Moes Moes that
uh entrench Monopoly power and so rather
than just thinking about how does a
monopoly in one market lead to another a
lot of the fact patterns that we're
confronting involve um how do some of
these technologies that you're talking
about some that you have experience with
have the potential not to just create
new markets of their own but to Rel
relieve and reduce the dependence on the
massive um tech companies or monopolies
um in the middle and so a lot of our
cases are about those flywheels those
Moes those feedback effects that might
insulate um those core monopolies from
competition and allow them to continue
to extract and stifle Innovation so what
you can expect throughout our
enforcement of these current cases but
also in the future are um the anti trust
division the doj uh my team fighting for
um uh competition to open up with
respect to those practices that deep in
the moat uh to allow those technological
disruptors to create um New
Opportunities and reduce dependence uh
and I think the cases that you'll see in
the future and the kinds of remedies
that we're going to demand um shall we
if we're successful we'll focus on on
that core um uh
Vision awesome thank you hi I'm Angela
and and I'm working on Andy it's an AI
search engine um I especially loved the
comments you said about regulating
monopolies and promoting competition and
my question for you is one thing we
noticed with Google is we think that
they have a browser Monopoly not a
search Monopoly anymore and so for AI I
think there's this huge potential to
unlock um more opportunities in
competition and how do you think that we
can prevent uh a monopoly from rising up
again when it comes to AI specifically
I think we have to study the past and
learn from it right and so I think part
of it goes to okay what are the the core
elements of a monopoly in the tech bace
and again this kind of goes to the
Eric's question and my answer which is
um it's more often than not the feedback
effect where is the inflection point
coming from and what are the assets that
a company might own in order to protect
those new competitors from emerging
right and so um you'll see for example
in some of our cases involving things
like sech that we talk about browsers um
that we talk about um the kinds of
advertising technology that might create
interoperability and Mobility for folks
who want to participate in multiple
ecosystems we talk about messaging and
other kinds of technologies that again
um allow people to Mo move cross
platforms and create the kind of um
technological um Innovation that we
believe is important so um that's kind
of how we're thinking about it but our
starting point in every case now is what
is the core Monopoly and does the
conduct insulate that core Monopoly from
legitimate um Innovative
competition uh hi my name is Jeremy
Nixon um I'm the founder of AGI house
which is this very charismatic you know
community of hackers and we all have
some complex beliefs about monopolies um
specifically Google brain uh where I
worked for three and a half years is
responsible for the creation of the
Transformer and of sequence to sequence
and of Bert and certainly with sort of
paid for with you know Monopoly money um
I would say similar story for Bell Labs
it seems like a lot of these monopolies
do sometimes use their resources for
enabling you know basic research and
technological progress and so do you
have a complex take on how to make sure
that that kind of thing is possible uh
similar story I guess um if Monopoly
profits aren't available or at least
like competitive profits aren't
available for a company like open aai to
take billions of dollars to create large
compute clusters just to have it be
Whitted away by meta relasing 305b um
how can we have VCS who will be
interested in investing in the creation
of the next Super cluster which creates
the next great model um very complex
questions but hopefully you have some
sense of yeah I mean no that's great
they're great questions so first let me
be very clear we are not against
companies succeeding and growing we just
believe that competition drives them to
do better the largest companies um can
you know innovate and bring great things
to Market and succeed and they do and I
don't see that go stopping anytime in
the near future um but what we do care
about is making sure that those
companies feel competitive pressure you
know to keep delivering um and and to
keep delivering in ways that um perhaps
uh might stretch them even further than
they think they can go so that's part of
what we're looking at second is return
on investment again is also something
that we think is um totally um
achievable and desirable and there are
plenty companies making a lot of money
um but business model competition is
important too right there are different
ways to monetize there are different
ways to um bring and and take value and
I think we want to let those um
Innovations compete with each other but
we also want to see those different
business models compete with each other
as well these are great questions F lot
fun thanks isi learner at jll um I
really appreciate the work you guys are
doing or you all are doing around you
know kind of breaking down some of these
Moes especially they involve business
practices but the other side of it is we
see a lot of I'd say certain firms using
litigation as a way to stifle
competition and really in a way that
they have Capital that's a startup does
not have and can really stifle
competition that way which isn't really
a a a technological moat but it is a
scale moat in the other direction yeah
listen um bigger companies get the more
aggressive they can get the more they
can try to intimidate right like a lot
of um the conduct that we see um
throughout the economy relating to
monopolies often resembles bullies and
Bullies can often throw lawsuits and
other things your way I think it's
important for us as um enforcers um
obviously there's a legal system people
have the right to use it we focus on
making sure that companies uh are um
investing to the greatest extent
possible and delivering results um for
the public uh and to us the more
competition they face um uh and the more
that um competitors can Thrive free from
impediments um the more that value will
be given back to the public hey um my
name is Joe dubet I'm a founder of a
company called Eden my question for you
is around the potential conflict between
competition and the consumer in terms of
protecting them I I see a lot of
instances where protecting competition
is totally in the consumer's best
interest you want as many pizza places
you can have in your neighborhood stuff
like that but in the situation where
like let's say a Craigslist or an
Instagram or something where there's a
benefit to having everyone in one place
how do you choose when you have to
choose between protecting competition
and maybe protecting consumer outcomes
sure I don't think that we have to
choose between the two I think we care
about CO as an anti trust enforcer the
animating um principle for me is that
competition yields great outcomes and I
didn't come up with that concept on my
own that's what Congress said when it
wrote the antitrust laws it said hey we
want competition why because we think
that yields better outcomes it promotes
opportunity it creates to a more uh free
and open and Democratic Society so um
that's the concept that we've been asked
to protect um and I I think when you
think about antitrust the work that we
do um it's not
saying you're too big or you're too
small or you should win or you should
lose it's simply saying hey you
shouldn't um engage in anti-competitive
conduct that makes it impossible for
someone else to compete if they have a
better idea or if they have a better
business model it says hey you know if
the dominant firm is going to gobble up
rivals in order to deepen its mode
rather than allowing some of those firms
to um continue in the market and succeed
and potentially go public or create new
opportunities then those could be
antitrust violations we're f F used not
on engineering outcomes we're simply
focused on addressing conduct that might
get in the way and I think that's kind
of how we think about it but I um do
want to be very clear I don't think
there's that trade-off I think good
healthy competition allows everybody to
benefit hi my name is Hansen and I'm the
founder of light sky and so you're
talking to a group of Founders um who've
come through I YC and you know obviously
we hope that we're the founder that gets
to IPO but a lot of our companies will
end in acquisition and so um there's
been a kind of chilling uh effect on
Acquisitions due to kind of the
investigations that the doj and the FDC
have kind of uh started recently and so
um can you comment a little bit about
kind of like whether or not like what
your perspective is on the chilling
effect of Investigations on Acquisitions
and how that so I appreciate the
question it's an important one um I
start with the data right we're all data
driven um the fact of the matter is the
overwhelmingly major majority of mergers
never get look uh single digits and then
even a small sliver of that actually get
challenged and so the deal economy is
moving deals are happening we only focus
on the deals that have problems and the
deals that have problems tend to be the
deals in concentrated Industries markets
or deals where you have a disruptive
rival who might be taking on an embedded
entrenched incumbent um and so those are
the kinds of transactions that we tend
to focus on um I think that um doesn't
impede exit it doesn't impede um the
deal economy in fact we see again plenty
of deals happening it's a narrow set of
deals that um satisfy a narrow set of
criteria that are the problem the other
thing and I think this is important is
um we want companies to have the
opportunities to go public we want
companies to build not just although
exit is a perfectly um normal way um uh
for companies to you know realize value
great world is one where companies can
build and they can find a path to
becoming public and um do so in a way
that is coste effective and can manage
um their resources as a public company
and then build to become thriving
durable businesses that can not only
increase the number of firms who are out
there um building new products and
services but frankly for startups
increases the number of viable acquirers
who can bid for assets and so what we
want to do is we get get this cycle
where there are only two or three firms
um sometimes one in each individual
space that is a potential acquirer
instead we want to encourage companies
um to build and to do that there needs
to be Pathways one um if you know to
sell to someone who's not an antitrust
problem and I think that happens all the
time but we wish it would happen more
because there are more companies that
wouldn't create antirust problems second
is we um want a world where where
companies feel comfortable that they can
pursue long-term Investments including
going public so that they can generate
value and grow and Thrive uh in a way
that is durable Jonathan it seems like
the the operating word that comes to
mind in like sort of a lot of the things
that you're talking about is like the
word open so you know on the foundation
side it seems like open source uh in
terms of models it seems like in terms
of public companies it's like you know a
public company is like an open cap table
I know you probably can't comment on
some of the Active cases but but you
know openness in terms of platforms and
being able to have comp competition at
platforms is like one of the things from
as an outsider it seems like that's you
know certainly a goal it's like
ultimately markets are about openness
whether it's a cap table the source code
you know or literally the platform that
allows other people to thrive
opportunity openness and opportunity are
are the foundations um for competition
and so we are really excited about
making sure that we can help preserve
that lot of our cases one of the other
themes is um you know a phenomenon that
I you know it's like open dominate close
right start open get dominant and then
you start closing aspects of your open
ecosystem so that others can't compete
effectively and I think making sure that
um the the availability of um folks who
are creating um who are dependent on
platforms and services who are um uh
induced to join and build on ecosystems
can realize the fruits of those
Investments and I think you know that's
that's what a healthy ecosystem look
like where where um where we have that
you know openness but I think that
you're exactly right Gary it's true in
in terms of having openness and
diversity of models business models but
also is uh openness and um to um to to
be able to succeed without having to
exit through acquisition and and um
openness of a public company that you
said like cap tables and um where where
the public can essentially participate
and benefit in um the building of these
great us
businesses one more quick one Lauren
good from W thank you so much for doing
this um I think what someone was
alluding to earlier was basically the
recent Manifesto from inre and Horowitz
about little Tech versus big Tech uh and
the idea that the regulatory agencies
are using Brute Force to
squash little Tech through m&a I think
you addressed that well but one of the
other things that was in the manifesto
was this idea that the government is
proposing attacks on unrealized gains
and I was wondering if you're able to
comment on that and generally what your
responses to this idea since there are a
lot of little tech people in the room
I'm just a small country antit trust
enforcer so I I'll focus on uh my little
uh area of the world which competition
policy and let tax experts focus on
those other things but when it comes to
making sure that um businesses including
startups have the ability to to thrive
and compete and grow uh and um that's
that's our focus when it comes to
thinking about the tech ecosystem and
I'll kind of finish where I started or
which is um you know I think you know
competition um is healthy competition is
a force m multiplier and strong
competitive markets can often uh
minimize um if not entirely eliminate
the need for invasive regulation and so
when we have an inflection point like we
have now it is the time to make sure
that we're doing a little bit of work on
the front end to keep our markets
competitive um so that um we can see as
many of our great business ideas succeed
and Thrive and stay in the market and
even um um become big open public
businesses thank you so much really
appreciate
it thank
[Applause]
you hi everybody I'm Luther low head of
public policy for YC so uh at YC we're
all about our Founders and so between
each of these policy conversations we're
going to have uh demos from our
community uh so I'm going to hand it
over to uh Mike n and saier we're going
to sort of feature how some of our
community is using AI all right hi
everyone uh first of all thank you Gary
and uh Luther for organizing this event
and uh inviting me to give a demo I'm
Mike I'm one of the co-founders uh and
the head of AI at zapier and my personal
mission is to pull forward the future so
that more people can experience the
extraordinary benefits of future
technology today zapier is workflow
automation software we are used by
millions of businesses across the United
States uh the vast major majority of our
customers are smbs so small mediumsized
businesses think like one to five uh
people companies and I find automation
can often be a little abstract to
explain so what I wanted to do is I
wanted to show you all a quick demo of
how zapier Automation and AI are working
together to serve many of the business
across the US uh let's imagine you're an
SB we're going to use like imagine
you're a coffee shop and you get
hundreds of customer emails every week
and your teams responding these emails
but you as the manager the owner you
want to try and get a summary what are
all what's a good summary of all the
things that people are writing about in
about so that you can Orient your
attention and spend time improving your
business I know we have a lot of policy
makers and representatives in the room
too so maybe in your world you can
imagine you're getting hundreds of
emails from constituents and you want to
try and get that summary of what are the
top things on everyone's mind so you can
make be better informed and make better
decisions about where you spend your
time so that's the problem and this is a
solution this is a zap it's what our
what we call our Automation and what I'm
going to do is I'm going to kick this
off by sending in an email to this like
fake uh email Dropbox that I set up for
this like fake coffee business uh hey
are you open during Memorial Day okay
and I'll hit send on this email and let
me talk through what zappier does here
so this is a zap that I set up uh ahead
of time and it's going zapper is going
to receive anybody emails that that uh
email it's going to add it to a
spreadsheet once we collect 100 emails
in that spreadsheet zapper is going to
release those and ask chat PT to create
a summary and extract out insights from
that and then it's going to email me
back a good summary of all of the things
that I've collected so far this is the
uh the spreadsheet that I've got I've
got a bunch of uh emails already in here
that are sort of fake and representative
and there we can can see my uh email
that I just sent got automatically added
to the bottom and if I flip back to my
inbox hopefully there we go just got the
summary back and this is a live summary
that was automatically generated using
zappier chiped mode back to about looks
like some order shipping issues product
quality concerns refund requests uh all
right I got a lot to lot to work on for
this business so this is a good example
this is just one of the millions of
things that folks use uh zapier for
zapier supports 7,000 Integrations on
our platform and our users plug-and
playay uh those Integrations to build
things that they care about one of the
things that I've learned is that over
the last couple years is that Ai and
automation I think are synonymous in a
lot of our users Minds um the promise of
both is software that just does more
work for you and I think this uh Insight
was what led us to go in all in very
early in AI in fact uh zappier went all
in on AI I personally did in the summer
of 2022 almost six months before chat PT
uh was released and zapier is running
about 10 million fully automated AI
tasks per month at this point I think
that zappier the largest fully automated
AI platform in the world at this point
and that usage has given us a front row
seat into what are what's the promise of
the technology what are people trying to
do with this stuff and also what are the
fundamental limitations that they're
running into and what I'm seeing is this
the number one problem right now is low
user trust that's due to low accuracy
and low reliability of language models
and it seems because we've been tracking
this for a few years now it seems those
problems are not going away with scale
and this was my first hint that the
underlying language model technology
seems like it might be inherently
limited now of course AI is getting
faster cheaper and we have more model
Choice than ever thanks to open source
but as I dug in I found something really
surprising the AGI Innovation
environment in 2024 is incredibly weak
and I think very few people realized
this we're stalling out on our progress
towards AGI and I think this is highly
surprising it was to me because big Tech
these AI Labs a lot of safety Labs
loudly promote this narrative that scale
is all we need to reach AGI just plug in
more training data make the models
bigger and we're going to get there but
this isn't true now I grew up in St
Louis Missouri uh famously the sh me
state so let me show you what I found
that led me to believe that this is a
chart of we put together of a bunch of
the most popular AI benchmarks out there
all the blue lines and you can see that
over time over the last couple years
progress towards human level performance
has been accelerating on these to reach
Human Performance what I found is that
the 's one eval one Benchmark called ark
and it's the only world's only AGI eval
that exists in contrast to all of the
other evals out there it's the only one
that actually measures AGI instead of
the more narrow form of AI this
Benchmark was introduced and created in
2019 on note before language models
before any of the Advent of scale that
we have today it remains unbeaten today
now you might be curious like okay
what's so special about this Benchmark
so here's an example this is uh one of
the 100 tasks or a representative
example of one of the 100 tasks that are
in The Benchmark and your goal is to try
and as a human figure out what's the
pattern between the task input and the
output and then map it to the test so in
this case you might you know see okay it
looks like we're trying to do you know
maybe fill in these uh blocks with the
square with the dark blue and we can
check our answer and there we go
confetti we got it right so uh this is
an example task these tasks are designed
to be incredibly straightforward and
simple for humans and yet empirically no
AI system can solve these 100 tasks
today this is incred in shocking to me
now Arc remains unbeaten despite for the
fact that two months ago I launched Arc
prize a million competition to anybody
who could get AI to beat this Benchmark
and so far no one has been able to
that's on top of the last five years of
evidence as well in the past AI
Innovation was driven through curiosity
through sharing through exploring new
ideas and instead today fueled in I
think part by big Tech and big aabs S
interest we now have scaling Dogma we
have closed Frontier research we have
llms as the only Paradigm Arc shows we
need new ideas to discover AGI I think
the industry's misrepresentation of
reality is distorting Decisions by not
only policy makers but Venture
capitalists by Founders by even students
there was 20 billion do invested in LM
startups in 2023 by my count only 100
million into AGI startups working on new
ideas half the students that I meet
don't want to work on AGI because
because they think it's a solved problem
and policy makers are even now
considering Innovation rate limiters
like
s1047 because the scaling Dogma is so
loud I want to repeat I think the AGI
Innovation environment we find oursel in
in 2024 is super weak the world is
basically betting that a single
commercial lab in isolation is going to
be the one to figure out and discover
AGI that is in direct contrast to how we
got here and why I'm even standing in
front of you we got here through open
progress open science and open sharing
that's not just true of AI That's True
of all science ever in fact I think if
I'd go as far as to say if you're
someone in the audience who maybe thinks
that we should pause AGI development or
stop AGI development entirely you should
be pretty happy with the world that you
find ourself in in 2024 but if you care
about reaping the benefits of AGI In Our
Lifetime like I do we need to work to
undistort this Market I think policy and
incentives should push towards open AGI
research and not rate limiting the
crappy version of AI that we have today
now I'm putting my money in time thank
[Applause]
you so I'm putting my time and personal
money where my mouth is I created Arc
prize with the express goal to provide a
very public measure of progress or in
reality lack thereof towards AGI and I
hope Arc prize can play a small role in
motivating researchers to work once
again on new ideas and openly share them
and help steer AI policy away from
regulatory capture and rate limits and
back towards open Innovation thank
[Applause]
you hi I'm uh Shawn Modi CEO of capital
AI uh summer batch YC um it's an honor
to be here among such esteemed guests
capital is a product to create
persuasive content from data in fact
it's so powerful some of Washington DC's
best lobbyists are already using it to
win multi-million dollar Appropriations
and create some pretty important
policies we're refining our product here
um at y combinator we already have
231,000 users and we are in the process
of finalizing a pretty large engagement
with Politico which I'll give you a
sneak peek of today so this is the nerve
center of capital uh under the hood
we're automating research so we're
looking at the entire open web and
pulling in the most trustworthy links
for the query at hand we're also
bringing open source and Frontier models
and a native document editing experience
when you click into this prompt box
you'll see the different types of
formats uh that you can select from the
art article is our most used format set
you can describe what audience you're
addressing you can say how long you want
it um we're optimizing right now for
quality of content not not speed so
you'll notice a little bit of lag as we
as we uh put in a prompt you can say
what sources you want to pull from here
Google search peer reviewed or have it
fully hallucinate and say none and then
you can select what content blocks you
want images web charts uh generated
charts metrics tables and and direct
quotes um so we'll put one in here and
let's
say uh Gary asked me to do one actually
so nothing like putting myself on the
spot so
sb147 pros and
cons let's let it rip
um so now we're gonna push this out it's
creating a headline for me understanding
sb147 pros and cons and it's going to
pull this information um in line why
while that's doing its thing I'll show
you some ones that I teed up already so
these you kind of get a cross-section of
my brain of what I'm curious about so I
learned about this company asml which is
they make these very powerful machines
that play a critical role in the uh
Global Supply Chain of semiconductors so
my prompt here was the geopolitical
risks of asml controlling key
infrastructure in the procurement of
semiconductors and you can see well
let's just jump back to sp10 here we go
so it's loading um you can see the
sources we're pulling from this is from
l in this is from gradient flow D Piper
it's dynamically generating a table so
let's just see sb147 also known as the
California safe and secure Innovation
for Frontier artificial int intelligence
models act it's a proposed state Bill
aimed at regulating development and
deployment of advanced AI models in
California this bill specifically
targets AI systems with significant
computing power particularly those
capable of Performing over 10 to 26
power operations interesting so if I
click on this paragraph I can go deeper
so unlike other chat Bots this is a
document creation experience so it's
making suggestions for me and it's
understanding the context of this
article explain the potential impact on
businesses and consumers that sounds
interesting so I'll let that prompt rip
and it's going to go pull from the open
web and and reprompt that specific
paragraph and it's asking me for a
clarification I'll do
that as federal
government cool um scrolling down we
have Dynamic tables generated here uh
looks like uh these are the different
provisions of the bill um it's pulling
out key metrics so 1026 operations
threshold for a models we don't
plagiarize when there's a quote we're
going to give attribution to where it's
from and you'll be able to see precisely
um The Source in this case this is for
Milan looks like it's from
Forbes and arguments against stifling
Innovation Exodus Talent impact on open
source economic impact it pulled in a
chart this isn't the sexiest chart but
we scan the entire page and we
understand the context of the page and
it's like having an analyst tell you the
chart illustrates the increasing share
of AI related research and development
expenditures at us colleges and
universities as well as the rise in
proportion of newly founded AI startups
among all Tech startups so on and so
forth and I can go deeper at the article
level here too convert this into a
dialogue between a policy maker and AI
developer discussing the pros and cons
long term uh economic Innovation impacts
I'll do
that so I'm quickly understanding this
topic and creating a piece of content
that could be useful for my job um and
then let's jump over here I'll show you
a few other things I've been looking at
uh I was curious of the Republicans
approach to FTC antitrust so I generated
a pretty in-depth report here of how the
Republicans and Democrats approach FTC
antitrust enforcement and if I want to
go and take that I'm let's say I'm a
journalist at Roy and I want to you know
break some news or have an opinion piece
I can go take this and put it into a
Google
doc and work with it and here it is with
all those tables embedded and citations
at the end of where the data came from a
few other things before I show you the
the political preview uh these are the
economic pitfalls of taxing unrealized
gains my prompt was write me a speech to
an audience of reporters startup
technology Founders and US Federal
regulators and why taxes on unrealized
gains is bad idea for United States
economy including lots of data so here
you can see excuse me uh paragraph one
there's seven sources that went into
this um draw linkable um if I wanted to
add a paragraph in between it
understands the context of the two
paragraphs and it'll make suggestions
analyze the implications for economic
growth and investment cool so I'll add
that in line if I need an image for my
article I'll convert this paragraph into
a synthetic image so I pass this
paragraph to a diffusion model and what
you're going to see is a a synthetic
image in line Looks like
this how I feel uh so um if I don't like
the styling of it in this case it's not
my favorite look um I'm going to change
it to something maybe more cyberpunk
where's my cyberpunk aesthetic let's let
that go there's my new paragraph above
with citations as we go down you can see
uh more tables so it's it's giving you
the information you need and in a fast
form factor um from very complex data
sources we're looking at 12 different
pages here that were fully scanned and
synthesized into a cohesive narrative
and fact checked um and uh yeah we'll
keep going here um I'll show you this is
there was a big incident yesterday in
Alaska of a joint Chinese and Russian
military exercise so I was curious about
that what are the what are the milit
what are the National Security
implications of this issue so I asked
for an in-depth analysis of why the
Chinese and Russian military fluid joint
exercise near Alaska um and I got a
pretty good assessment um of um the
history of their joint exercises and
what it means for the
Arctic uh I went deeper natural resource
context of China uh and Russia interest
in the Arctic um so you get you get the
sense so now zooming out you can all use
Capital today CIT l. a uh for the
affordable price of $8 a month uh that's
if you use the discount code design
Capital with a no but now I'm going to
show you something that's up and coming
and pretty exciting so we are friends of
Publishers you know I live in Washington
DC some of my closest friends are
journalists and I think right now ai has
a kind of you know hands keep me at
hands distance um relationship um we
want journalists to thrive we cannot
live in a world without journalists
journalists play a critical part of
democracy so that's why Politico reached
out to us uh because they want to embed
AI into their political Pro product um
pretty much every major lobbyist every
major government official every
Congressional office every senate senate
office uses political Pro to understand
the inner workings of Washington DC uh
so this is their product and what you'll
notice is a little widget called AI
reports powered by Capital AI so if you
are curious of how Congressman R con
versus Blake Moore voted on Ukraine Aid
um or on Israel Aid or on other types of
key matters you can put in that prompt
and it would look at all the political
Pro proprietary data and give you an
interactive report so I'll give you a
preview what that actually looks like if
you go to the open left nav you'll see
create an AI report we're going to
introduce this concept of projects so
you can add multiple URLs articles data
sources to this to basically a bucket a
folder this case this is 2024 funding
bills I'm going to add those sources to
my prompt box and then it's going to
suggest different prompts of how I can
interrogate that data and write a report
so I'm going to go with the suggested
version version write an article
explaining Bill
s205 and its impact on the interior and
EPA and just like you saw in our open
product you're seeing an in-depth
analysis of this bill and this was
created in capital obviously this
prototype for for but superposed for
political Pro this will be a full
document editor experience where you can
edit text highlight text and add
specific data to sections you can add
YouTube videos PDFs word docs
legislative data regulatory data and
prompt it um so here you can ask Capital
to provide examples of specific
provisions of the bill or you can take a
testimony from Court add it and say
highlight the key Parts where you know X
happened but it doesn't end just there
so X out at the document level this is
where you'll manage all your sources you
can remove sources add them regenerate
things and if I want to turn this into a
tweet storm or a piece of content to
influence policy not only understand
policy but to influence policy I'm going
to go turn this into a tweet storm
and then I'm going to prompt it saying
write a series of Tweets in opposition
to this bill targeting
spending and what I have here is a tweet
storm opposing this bill citing articles
from political Pro and other sources
that are relevant and if I click that I
can reprompt it using AI so the cycle
continues so that's a little preview of
shapeshifting as we call it with uh with
our product you can see that original
Source here how AI will be infused
throughout the entire analysis process
this goes into the regulatory aspects of
well as well um your average citizen
it's hard to understand the legal ease
of many of these things and we want to
democratize that so yeah that's what
we're building Capi open to your
questions afterwards and thank you for
your
time thank you
Sean up next we have FTC chair lenina
Khan is going to make some remarks and
then uh she's going to participate in a
fir side chat similar to How Gary did
with Jonathan we've flown out Timothy B
Lee does anybody uh
read understanding AI outstanding uh
substack I highly recommend it it's
really helped uh a non-technical person
like me kind of understand what's going
on under the hood uh and um so I'm just
going to quickly uh introduce Timothy
and cheran cheran will come up and speak
and then uh Timothy will come up and
interview her so and we'll have some
time for uh Q&A for the audience so uh
Timothy is a technology journalist and
analyst who writes uh the popular
substack understand AI uh with over a
decade of experience covering Tech
economics public policy he brings deep
expertise to his analysis of artificial
intelligence and its impacts uh he
previously wrote for Arts Technica Fox
Washington Post he has a master's degree
in computer science from Princeton um
and through his writing he really helps
readers grasp how AI works and how it's
reshaping our world um his newsletter is
exploring topics like large language
models self-driving cars ai's effects on
labor markets
maybe SP 1047 I don't know maybe we
maybe we'll see an article about that uh
soon uh the philosophical questions
raised by artificial intelligence and
then uh Lena M Khan is the chair of the
Federal Trade Commission appointed by
President Biden in 2021 at the age of 32
you know this is Lena Khan's thirdd YC
event we had uh my first I was like two
weeks into the job and we had her come
out uh last November and then she came
out or we did a policy event in
Washington called remedy Fest and Not
only was Sher Khan there but we also had
Elizabeth Warren and JD Vance so it was
like this interesting uh assembly and
then at the event JD Vance said my
favorite member of the Biden
Administration is lenina Con and so now
that he's been named uh the vice
presidential pick on on the sort of
trump BS ticket everybody is like you
know scouring the the transcripts of uh
our first real policy event in
Washington remedy Fest so that was kind
of a cool thing that happened last week
anyway sorry I got got distracted uh
legal schol a legal scholar and
Anti-Trust expert con Rose to prominence
when her 2017 Yale Law Journal Amazon's
antitrust Paradox which challenged
traditional antitrust thinking uh prior
to her FTC role she was associate
professor at Columbia Law School served
as Council to US House Judiciary
Committee subcommittee on antitrust
she's known for her Innovative approach
to competition policy advocating for
more aggressive enforcement against Tech
Giants and other large corporations her
Works uh reshape the conversation around
Monopoly power and the digital age
earning her recognition from
Publications like Time Magazine foreign
policy she has a JD from y Law And a ba
from Williams College please welcome to
the stage cheer
[Applause]
con hey everybody it's so great to be
here as Luther mentioned this is my
third YC event and before each one
Luther really tries to convinc me to
ditch the Blazer and just wear a t-shirt
so I can fit in but I haven't been able
to get there so I hope you'll forgive uh
the Blazer though now I see Luther is
also wearing a Blazer so I don't know
what what happened
there um well it's so great to to be
here and as you all know there's been a
lot of talk recently in Washington about
how to address Ai and Gary has been just
at the Forefront of making sure that
little Tech has a seat at the table uh
and so that to make sure that we are
hearing from Market participants across
the board and so really appreciate that
that advocacy and um championing the
role of little Tech and so against this
backdrop it's really exciting to be here
to get to talk about how we can make
sure that America can harness the full
opportunity Innovation and growth that
AI could present and I believe that a
vital ingredient to making sure that
these markets stay open honest and
competitive is going to be vigorous
antitrust enforcement and competition
policy Fair competition is what ensures
that the best ideas win that markets are
actually rewarding businesses with the
best skill and talent rather than
rewarding just the businesses that can
exploit special privileges or advantages
and these are really the markets that
allow tinkerers and dreamers to come up
with a big idea take a risk and
Thrive historically uh these open fair
and competitive markets have been the
engines of American growth and
innovation especially at moments of
major technological disruption and
transformation uh like the one that
we're seeing now with AI and so these
are the moments that we in particular
see breakthrough ideas ideas that can
really shift the Paradigm but history
shows that these are also the moments
when incumbents can try to tighten their
grasp even if it means abusing their
power because they have the most to lose
so these companies can potentially
exploit bottlenecks and stop the flow of
innovation uh they can leverage their
Market power to pick winners and losers
rather than allowing the best ideas to
win and so to make sure we fully harness
the benefits and opportunities of AI we
have to make sure that we're staying
Vigilant and that we're making sure that
upstarts can really compete on a Level
Playing Field here and I believe that
this is going to require alongside other
things a real commitment to a philosophy
of openness across the industry uh which
means open markets but also open
architecture open ecosystems and open
source software and so I'm going to
share a little bit about how we at the
FTC have been thinking about that for
decades open open source software has
driven competition Innovation and
opportunity in the tech space under
certain conditions it has allowed
researchers and developers to build on
each other's discoveries more easily and
more efficiently take Linux for example
which was developed more than 30 years
ago with the help of the community that
sprouted up around it it has allowed
countless Technologies to flourish from
cloud services to supercomputers and now
it Powers some of the world's most
important systems from the New York
Stock Exchange to the particle
accelerator
atern it would not be an exaggeration to
say that nearly all of Y combinators
most successful companies would probably
not even exist without open source
software and it's community so many of
the technologies that we use every day
would still be just ideas because the
barriers to entry posed by proprietary
software can sometimes just be too high
openness is more than just a feel-good
philosophy it's been a proven Catalyst
of innovation which is why it has
attracted hundreds of billions of
dollars in Venture Capital funding to
help startup Founders bring their ideas
to life so it's worth thinking about
what open source could mean in the
context of AI I both for you all as
innovators but also for us as law
enforcers of course the definition of
Open Source in the context of software
does not neatly translate into the
context of AI and I know there's a lot
of discussion still figuring that this
out and I've been really grateful to get
to chat with some of the leaders um that
are really thinking through what a
shared understanding of openness could
mean in the context of AI as a starting
point instead of saying open sour models
the FTC has been using the phrase open
weights models specifically referring to
AI models for which the weights are
publicly available it's still early days
but we can already see that open weights
models have the potential to drive
Innovation promote competition and
consumer choice and lower costs and
barriers to entry for startups like the
ones that incubate here the FTC is clear
eyed about the conditions that need to
exist to make this Vision come true as
you know better than anybody it isn't
easy to build an AI Foundation model
it's resource and capital intensive from
retaining engineering talent to
accessing expensive computer in uh
compute infrastructure to acquiring the
necessary
data these conditions have allowed in
some instances the biggest technology
companies to get a leg up in the AI race
if you control the raw materials you can
control the market and shut out smaller
companies who don't have the
infrastructure to compete this can lead
to fewer exciting products made by even
fewer companies and can come at the
expense of both Innovation and consumer
choice but with open weights models more
smaller players can bring their ideas to
Market there is tremendo potential
tremendous potential for open waight
models to promote competition across the
AI stack and by extension spur
Innovation across the stack too open
weights models can reduce costs for
developers so that they can focus their
capital on products and services rather
than expensive model training and they
can free up Venture capitalists to
pursue promising new applications of
models rather than starting at square
one with model
development so at a basic level open
weights models can liberate startups
from the arbitrary whims of closed
developers and Cloud
Gatekeepers developers that use open
models we've heard can feel like they're
less likely to have the ground shift
under them because one day model owner
decides to significantly increase the
cost of API access this is the type of
free and Fair competition that the FTC
is committed to promoting you all
deserve the opportunity to build in that
type of environment free from the undue
sway of Monopoly power of course open
weights models can also come with some
challenges and risks first it matters
Who develops and owns the open model
we've seen firms deploy opportunistic
open first Clos later strategies where
they use openness to draw in developers
scale up quickly and enjoy the network
effects and data feedback loops that
this scale provides then once they've
ridden openness to dominance and locked
in a user base they can flip the switch
and become closed instead policy makers
across government need to be V Vigilant
of this Playbook and antitrust enforcers
already are for example in one of the
ftc's lawsuits against meta we alleged
that meta in Web 2.0 had initially
allowed third-party developers to build
apps that integrated with Facebook only
to reverse course later and restrict
access to those that challenge them the
Justice Department's lawsuits against
Google allege a version of this story as
well second the licensing terms attached
to the model are crucial a developer
could make a model's weights available
under licensing terms that ultimately
restrict developers from using it to
meaningfully compete in the marketplace
and third there's a serious risk of Bad
actors using open models for deeply
concerning activities AI can be used to
clone voices to defraud consumers and to
create sexually explicit imagery of
people without their consent including
children and this isn't speculation we
are already seeing it and so even as we
Embrace a commitment to openness we need
to be clear right about these risks
openness is just one way to promote a
Level Playing Field for startups there
are other ways that we can shape policy
to structure markets so that they really
promote free and Fair competition that
allows the best ideas to rise to the top
Founders have told us that they struggle
to compete because dominant players may
be monopolizing access to great talent
to critical inputs and to valuable data
the FTC is doing our part to be vigilant
and to open up the market and ensure
that Founders have what they need to
start and scale their businesses first
talent some of the best engineers in
America have been bound by restrictive
non-compete clauses we've heard from
startups that secured funding and
entered the market only to find that
they can't grow because the talent pool
has been locked up by the dominant
players earlier the this year the FTC
banned non-compete clauses this will
allow 30 million Americans including
developers designers and researchers
across the country to move freely from
company to company with their innovative
ideas and unique
[Applause]
expertise second we're making sure that
you all have access to the critical
inputs to build AI tools and models one
of the first merger lawsuits that we
filed after I joined the agency was to
block nvidia's attempted acquisition of
arm which would have given one of the
largest chip companies control over the
technology and designs that its
competitors rely on to develop their
their own ships our team determined that
the merger would undermine competition
and hamstring innovation of Next
Generation
Technologies we also launch an inquiry
into the Partnerships between dominant
AI PL players and Cloud spe uh and cloud
service providers to better understand
the impact of these relationships on
competition and to make sure that no
company is exerting undue influence or
gaining special access because of the
other firms that they happen to work
with
third we're making clear that major
companies cannot collect data by
surreptitiously changing their terms of
service this is not only an invasion of
consumers privacy but it can also
distort the playing field and give these
firms a leg up over smaller competitors
that have fewer avenues for data
collection at the helm of all of this
work is the ftc's new office of
technology which we launched last year
to deepen our in-house expertise
our team includes folks who have built
open-source software and who have
deployed this technology to millions of
Americans we have active members of the
open source community and alumni of
startups that incubated right here at YC
I want to close by addressing what I see
is a common misconception about the tech
sector in policy circles which is that
Tech is a monolith that the interest and
incentives of all companies are the same
be it the scrappy startup or a giant
firm the many conversations that I've
had with Founders in Silicon Valley have
really underscored just what a
misconception this is uh we've heard a
lot of excitement from Founders about
the potential of this moment but also
some weariness of the existing control
and influence that some of the
incumbents have and so it's really just
underscored uh the importance of us in
DC to be hearing from a broad swath of
Market participants uh little Tech as
well as the big guys and so I'm really
grateful uh to you all for hosting me
and really excited for the conversation
thank you so
much hi everybody um I'm timly I'm I
read a newsletter called understanding
AI as Luther um mentioned earlier um
thank you chair Conan for coming um so
uh I feel like the classic um and in
trust case is like a company that's been
around for decades and is you behaving
anti competitively and it's like clearly
a monopoly um and in with um large
language models we have kind of the
opposite where you have this U Market
that basically didn't exist at all two
years ago and we went from kind of
nothing to open eye has billions of
dollars to revenue um can you walk me
through how the FDC thinks about a
market like that because it seems like
it's pretty Dynamic right now but you
could certainly imagine a company you
know coming up in a monopolistic
position what do you kind of walk watch
for to see like whether there's anything
the have Toc or any regulars more
generally should be doing or whether
it's just kind of kind of stay back and
see how the market develops yeah it's a
really good question and and one that's
really informed by the experience of the
last few decades and so in the early
2000s we similarly had a moment where
you saw a lot of you know dynamism uh
among a set of companies and there was a
sense in DC that the best thing for the
government to do including in the
context of antitrust was to just step
back and entirely get out of the way the
thesis being that these markets are so
fast moving that even if you have
dominant players even if you have
Monopoly power that will just naturally
be eroded in the marketplace and so that
was kind of the governing thesis you saw
a whole set of Acquisitions uh be
allowed to go through hundreds of
acquisitions by you know the four or
five big tech companies and you know
fast forwarding today I think we
realized that aspects of that thesis
were really misguided and really under
indexed the significant barriers to
entry that you see in digital markets I
think there had been an assumption that
you know people don't have to build
massive factories to enter these markets
and so uh we're just going to see a lot
of Entry that will discipline any
Monopoly power instead what we saw was
that you know Network externalities the
feedback loops of data really tended to
lead some of these markets to tip such
as that you know early players and
players that were able to get a leg up
and really deepen their Moes create
entry barriers raise those entry
barriers could end up actually
solidifying dominance and Monopoly power
in a more long-term way so that even
once they started acting in ways that
weren't in the best interest of their
customers be it on the business side or
the user side it became much more
difficult for entrance to come in and
discipline that and so over the last you
know few years including under the last
Administration we've seen major lawsuits
filed um a subset of those lawsuits are
trying to correct for some of the
inaction of the past especially with
regards to some of these Acquisitions
but also with regards to some of what we
now realize was anti-competitive conduct
and so that experience has been a
cautionary Tale in terms of the
downsides of just assuming that these
markets are so Dynamic they're so fast
moving so we don't have to worry about
Monopoly power at all of course the
specific ways and tactics that dominant
firms might use today in the AI context
and in the llm context will look
different but I think we've learned that
there can be real downsides to not
addressing Monopoly power if it's being
created or used illegally early and
instead allowing it to deepen and become
much more difficult to fix on the back
end the other thing that we really lose
is a lot of innovation right I mean when
you have monopolists unlawfully be able
to squash some of these innovators or
create a chilling effect so that some of
these ideas don't even come to Market a
remedy that you might get 5 years down
the line or six years down the line can
really never fully make up for all of
that Innovation that was lost and so
that's why we think it's especially
important to try to prevent this stuff
on the front end rather than playing
catch up and cleanup on the back end
when you've already lost so much
Innovation and
opportunity um so I know that in
addition to regulating antitrust you're
also the FTC is a major um uh agency in
terms of privacy regulation and um I
know one of the one of the um concerns a
lot of companies have and and um
consumers about about large language
models is um you know sending their data
to a um to a big company like Google or
open AI um are there um privacy benefits
to open we models um in terms of being
able to run these models on device or
you know on a company's own um premises
as opposed to um sending them out to
some to some third third party it's
possible I mean I know we've seen some
of those claims be made and I think we'd
want to you know take a closer look
before are clearly saying that you know
there are definitive privacy benefits
here um but those are absolutely
concerns we've heard from consumers but
also like Enterprise customers who will
say you know sometimes we are using some
of these models and we don't have entire
confidence that our proprietary
information is not being fed back in and
we think the terms of service are
written in a way that would protect us
from that but I'm not sure and those
terms of service actually changed last
week so maybe it used to be safe but now
isn't and just just a general
uncertainty about what could be a pretty
core issue for a lot of firms is whether
they're competitively sensitive
information and proprietary information
is or is not being fed back in so we've
been putting out um kind of notices and
blog posts for the market to just put
firms on notice about what types of
practices could be illegal under our
consumer protection authorities and so
the consumer protection laws prohibit
conduct that is deceptive and so if
companies are representing one set of
data practices but actually engaging in
a different set of data practices that
could be deceptive you could also just
deceive people by omitting certain key
information and so we think how this
information that's being fed into these
llms as being used is a material term
and so we've made clear that um
companies need to be upfront about that
so we've had several people this morning
mention SP 1047 um the California um
legislation on safety and one of the big
criticisms of that is that it might
discourage companies from uh releasing
openweight models because it places a
lot of um requirements on companies that
release those openway models um is that
something that you're concerned about I
mean not going to weigh in on this
specific bill but as a general matter I
think it's it can be tricky right
knowing how to strike the right balance
and I think candidly policy makers do
have a lot of legitimate concerns about
how some of these tools could be misused
and they're already seeing some of that
play out in real time and so they feel
an obligation to the public and the
constituents to protect them from some
of these risks that said I think it
would be a real missed opportunity if we
didn't position openness to really get a
fair shot in this Marketplace especially
from a competition perspective where
we've seen that in this type of Market
where you know the entry cost can be
quite prohibitive openness can be
especially important because it just
lowers those entry costs it kind of can
Focus people on um you know innovating
in more efficient places and so you know
I I do think we need to craft policy in
a way that preserves the opportunity for
that openness in in your remarks you
mentioned um the risk of fraud from Deep
fakes and uh AI um I was wonder if you
could tell me a little bit more about
what you're seeing I me this is
definitely something people always talk
about but you're probably in a unque
position actually see kind of the
magnitude of this is this still like a
mostly theoretical thing where it's a
few places or something that's really
happening on a large scale so the FTC
does um police fraud and we you know for
years now one of the most common sources
of fraud that we see in our complaint
database is impostor fraud so people
pretending to be the government or
pretending to be a legitimate business
uh calling up people pretending to be
the IRS saying hey you owe thousands of
dollars and if you don't you know wire
over money in the next 24 hours we're
going to come and arrest you like these
types of scams are actually still quite
prominent um what we've seen with fraud
with um some of these AI tools is that
they risk turbocharging these existing
types of fraud because it allows
scammers to disseminate fraud uh much
more quickly much more cheaply on a much
wider scale with voice cloning Tools in
particular we're also seeing an update
in complaints of you know people
pretending to be somebody's grandkid and
you know figing their voice quite
effectively um and so we've been
thinking through you know how can we be
most effective one thing we recently did
was uh launched a voice cloning
challenge where we actually invited the
public to submit ideas to us about how
either us as a law enforcer or people in
the public could potentially detect
voice cloning fraud in real time and so
are there Technologies or ideas already
out there that would allow you when
you're getting a call to figure out in
real time is this a real voice or is
this a fake voice we got some really
interesting ideas and just announced a
couple of months ago um the three
winners um and so you know we
spotlighted those we hope that will help
them you know get more uptick in the
market and really um be able to scale
and so uh you know we're using all law
enforcement tools but we're also
thinking more creatively about how else
can we encourage Innovation not just in
fraud but innovation in fighting and
detecting fraud questions sh in the back
thank you um Greg Miller building 80 and
AI legislative affairs platform curious
to know we see a lot of arguments that
some of these firms are so big that even
the big firms need to engage in
anti-competitive Behavior or
Acquisitions to compete against the
other big firms we see this a lot in the
flight industry airline industry we also
saw you know Google needs to pay Apple
202 billion dollars and then Apple can
compete on other fronts right I'm
wondering how you think about that how
you think about these companies being so
large that they need to do certain
things to compete against the other
really large companies uh so as a
general matter we never think illegal
practices are okay even if other people
are doing them
um um you know it it it's it raises an
interesting point and one we definitely
hear in all sorts of context uh you know
famously in healthcare I think we've
seen a lot of consolidation and you know
some of these markets can get into a
mode where you start seeing an arms race
of companies that are having to bulk up
and they say we're having to bulk up
because the other side bulked up and you
know it's kind of ends up being uh this
race to
monopolization and you know we're law
enforcer so we look at what do the laws
say and the laws say actually you need
to be able to stop consolidation and
incipiency right Congress identified
Fair competition is the way that we need
to structure our markets to make sure
that the public can benefit be it
consumer small businesses workers and so
we have a mandate to actually crack down
on some of that I can certainly see that
if you have you know it years of
inaction on the law enforcement front
companies can feel well I'm not being
protected from this Monopoly abuse and
so my only way to survive is to become a
monopoly myself and you know I think
that really underscores the importance
of enforcing the law so that firms are
not put in that position um but I think
the risks of consolidation and unlawful
monopolization can be so high uh that
it's really not a good idea to double
down on that um I mean in addition to
the harms that we've been talking about
I think we're also seeing the resiliency
cost
that can come with centralization right
if you're centralizing production you
can also be centralizing risk so that
when you have a single disruption be it
a natural disaster or you know a
contamination in a factory that can end
up having cascading effects and so the
whole system breaks down and so I think
uh you know the the resiliency risks The
Innovation risks um of some of the
centralization are so high that I don't
think we we want to go
there hi hi I'm Emily founder of clearly
AI we're doing security and privacy from
first principles so I had a question
around your uh role to protect against
deceptive practices in consumer privacy
specifically around how you're thinking
about the fact that today it is a
default opin world for training on your
data and specifically thinking around
Slack's announcement to default opt-in
everyone to train on your slack
conversations and you have to email an
archaic email in order to opt out how is
the FTC thinking about that versus
switching over to a default opt out
world yeah it's a really really
important question and one that we've
been grappling with um a few months ago
we actually did a round table with a
bunch of creators uh people across the
creative professions from authors to
graphic artists to fashion models um
people who have seen that you know they
wake up one day and overnight their
life's work has been by default fed into
some of these models and in some cases
is now being used to compete with them
without their permission and so one of
the big issues that they flagged is that
all of this has been an opt out model
rather than an opt-in model in addition
to you know the lack of compensation and
that that they think fundamentally needs
to change um we've put out some notice
for for firms about instances in which
overnight changing your privacy policies
so that by default firms are opted into
that or people are opted into that
scraping um could potentially be illegal
in some cases you really can't mislead
people you shouldn't make it
extraordinarily difficult to exercise
the choice to actually opt out um and so
that's something we're we're tracking
closely hi there my name is nikelle I'm
founder of a company called rescript in
the current batch uh I'm wondering if
you could talk a little bit about um the
overturning of the chevron case and what
that means for companies who have to now
I don't know comply with the different
sort of regulatory environment good
question I wasn't expecting a a supreme
Court uh Juris Prudence question here
but I I love it um you know this is
definitely a moment where we are seeing
courts uh revisit some you know key
principles that had been in place for
many decades and so uh you know it's
it's been a moment of uncertainty and
some destabilization even for government
agencies as we wait to see how the court
rules on some pretty foundational
questions you know in the Chevron
context in particular the court has
basically said that even when a statute
is ambiguous um the courts should not
necessarily defer to the agency's
interpretation of that ambiguity um and
the courts really are the ones that
should be in the driver's seat and
figuring out what the law says and means
so you know generally there's much more
um skepticism of administrative agencies
and kind of just deferring to that
expertise um in practice you know uh
importantly though the court also said
that although we are overturning Chevron
that doesn't mean it's open season to
revisit all of these previous decisions
that relied on Chevron so they're kind
of limiting the potential
destabilization of of of that um but
it's something you know we're following
closely uh we think everything we're
doing is already in compliance with
these uh with these new rulings but it's
definitely a evolving landscape um there
going to be some you know big decisions
teed up for the court next term as well
that we're going to be following closely
and so uh you have to kind of stay
Nimble and and see where things land
good afternoon thank you very much for
sharing your insights we're not a
Founder I'm I'm representing the
European Union here on the US West Coast
my name is k so we're innovating a lot
in regulation but not unfortunately so
much in Ai and and these other things
but I I mean my question and thank you
for your leadership and also for the
very close cooperation between the FDC
and particularly I mean DG competition
and increasingly also DG connect I mean
you mentioned the philosophy openness
you also said it can take time and then
there's four on like Innovation and I
mean we had the same analysis in the
European commission European union and
we came to the conclusion we needed an
additional tool in our toolbox an exante
tool and we couldn't just rely on expost
because it was taking sometimes too long
and it was Case by case and by the time
that the case was resolved the
challenges was were dead and so we have
the digital markets act it's still
relatively recent it's a couple of
months four five months but my question
to you is do you think the US should
kind of needs a dit markets act would
you be in favor of a digital markets Act
of exante to give yourself more margin
of
maneuver I mean I defer to Congress on
on those issues they're really the the
key players to figure out what
additional laws are needed we're really
charged with using the existing laws and
tools that exist um it is a very you
know interesting moment for ideas about
what to do here if anything one concern
and question that I've heard from some
including in the tech world is when you
have companies that are vertically
integrated across the entire stack
including with layers that are essential
inputs for Downstream players does that
create a core conflict of interest like
if you are dependent on a firm that is
also now becoming your biggest
competitor like how does competition
really work there and I think that's an
important question um historically there
have been markets in the United States
where lawmakers said there have to be
some structural separations and so if
you own the railroads you cannot also
own the coal mines because there's going
to be a risk that the railroad will only
transport their own coal and not of The
Independents and then The Independents
have nowhere to go right similarly in
our banking laws we traditionally had a
system of separating banking from
Commerce because you didn't want the
companies that control the credit Supply
to be competing with all the companies
that depend on them for the credit
Supply so historically we have had
markets where we recognize that certain
players played such a foundational role
for Downstream economic activity that we
couldn't risk corrupting the incentives
up top we really needed to keep them
neutral in a certain way and so you know
that's an idea that I've heard in in
various discussions and it that would be
clearly something that would need to be
done through um new laws and so it'll be
interesting to see what
happens uh hi uh my name is Ben I'm the
founder of a company called K scale Labs
Building open source humanoid robots um
I was wondering about um basically this
is more like like soft question but um
there's a sense that like I feel like a
lot of government agencies are getting
more politicized and I was wondering you
know I I kind of view FDC is a very
important um sort of pillar of tech
Innovation I was wondering how you think
about that like what's it like in DC
like how do you make sure FDC doesn't
get like
defanged it's it's a really interesting
time in DC because we are seeing as
you're noting a lot of kind of
disagreement on really major economic
and social issues but on issues around
antitrust and anti Monopoly we are
seeing enormous bipartisan agreement
that we really need to double down on
Fair open and competition uh competitive
markets we can't just seed markets to
monopolies that can pick and choose
who's winning who's losing uh and you
know we see some of these concerns in
the tech context but we also see them in
the context of our agriculture markets
our Healthcare markets uh one of the
biggest areas of bipartisan agreement
with the ftc's work uh is concern about
whether dominant players in the
pharmaceutical supply chain could be
using their power in ways that's
inflating drug prices for Americans but
also squeezing Independent Pharmacies
and hollowing out parts of Rural America
that depend on these pharmacies for
essential health care so there are a
whole set of issues where you know
people that represent all these
different parts of the country are
seeing firsthand the costs and harms of
allowing concentrated economic power to
be unchecked to really you know be
exercised and anti-competitive ways and
so um you know that that bipartisanship
and agreement across the aisle of the
importance of the work that agencies
like the FDC do are are really important
um you know that said uh the the
monopolies and and dominant firms that
the FTC is suing and bringing actions
against um they have power and clout and
so you know that type of push back is
something that we certainly see and it's
kind of embedded in the DNA of of an
antimonopoly agency that when you are
doing your job you're going to see some
of that push back but uh it's really
important to kind of stay focused and
remember that we serve the public and so
you know that's where we need to keep
our attention last one uh thank you for
being here and and thank you Gary and
Luther and YC for organizing this my
name is Dave Yen my question um you know
it's great to hear the focus around open
source as well as specifically open
weights on the models um but one of the
biggest limiting factors for the
acceleration of AI is access to to data
and high volume high quality data and so
what what are the ftc's views on um data
sharing broadly as far as uh helping to
advance AI with respect to Consumer
privacy as well yeah I mean I don't
think we have kind of a dogmatic view on
that I think from a data a privacy
perspective we always want to make sure
that people's expectations aren't being
upset that they're not being deceived or
tricked about how their data is being
used um I'd be really interested to hear
you know what are some of the ideas that
you all have about if if access to data
is right now a major barrier to entry or
to to scaling you know what are ways to
address that uh and what would it look
like to make some of this more available
but in a way that's not you know
jeopardizing people's privacy um so you
know we haven't come out with a fixed
view on that with the exception of
wanting to make sure that if data is
being shared it's not being done in a
way that's deceptive or unfair we've
also been very clear that there's
certain categories of data that are just
especially sensitive right so People's
Health Data people's browsing history
people's precise geolocation data and so
for those categories of data there is a
heightened obligation that firms have to
not you know surreptitiously share it or
sell this data without getting
affirmative permission from consumers
just because that type of information is
especially
sensitive give it up for Timothy and L
con thank
you we we now have an another Lena Lena
kuchi who's one of the most impressive
Founders and impressive companies from
our winter 24 batch is going to do a
five to seven minute demo of infinity
that you're going to love
great hello hello everyone uh my name is
also Lena and I'm also wearing a
Blazer um so I'm really excited to talk
to you today we are Infinity AI um we
are building a script to video
Foundation model and uh basically the
way it works is users type out a script
they say they write out what they want
their characters to say and then they
get a video out
um and we train all our own video models
and so I wanted to give you a demo of
how that works today all right so we can
either type out a script from scratch or
we can similar to chat jpt use a prompt
and kind of get this first draft of a
script so I'm going to do that I'm going
to use an AI
prompt and let's do YC Partners complain
about Founders forgetting their badge
you're part of YC that's probably a
common occurrence these days um and so
when we generate the script we'll get
basically a first draft here we have
Gary saying how many times can we remind
them bring your badge so be saying
seriously it's a small thing um so if we
want we can edit this script we can
regenerate it until we like the way it
looks um and then we can press
generate um and you'll see there's kind
of a a new pop up here and this will
take a few minutes to generate and so in
the interest of time I uh already
generated a video with uh this prompt
right before this and so let's watch
this
video another day and another founder
forgetting their badge it's
unbelievable I feel bad for their
employees how are they supposed to run a
company if they can't even remember
their badge do we think they're allergic
to remembering serby are there allergy
kits we can get for
that so this is the editor page um and
so here we can go in and change up the
script regenerate Clips change things
around and then once we're happy
restitch things together and then share
out our video um and so you can kind of
get the idea here of this like script to
video experience I think for us you know
the main thing that motivates us at
Infinity is
storytelling and our core Insight back
in the day was people are at the center
of stories and so if you have expressive
characters if you have good characters
you can tell really good stories even if
you don't have anything else and so
that's kind of the focus on characters
and the script and the dialogue driven
nature of the product and of the video
models that we
train um one of my favorite things to do
with our tool is actually to imagine
people from the past and characters from
the past from the era before they were
video cameras and um so specifically for
this event uh I made a video yesterday
and it features people um from you know
luminaries from arts history um and so I
wanted to play that I think it's very
relevant to the types of conversations
that we're having at the event today um
and it's also a reminder that the
history of art has always been very much
linked to the history of
[Music]
Technology printed books will never be
the equivalent of handwritten
codices the camera will never compete
with the brush the gramophone will lead
to a deterioration of American music the
cinema is little more than a fat what
audiences really want to see is flesh
and blood on the
stage people will soon get tired of
staring at a plywood box every
night who the hell wants to hear actors
talk we are Infinity thank you so
much in Washington sort of an important
part of the advocacy ecosystem are civil
society or nonprofit uh
organizations and uh that's been
especially important in the last uh
decade or so when it comes to technology
policy and so I'm really excited not
only to sort of have these folks on the
stage but really to partner with them on
this event um uh so I'm just going to
have you all introduce yourselves uh and
uh tell you tell us about where you're
from and just very quickly because we're
going to speedrun this panel to get us
back on uh schedule um what your
organization is doing and how it's
thinking about open source AI so hi I'm
Haley skama I'm associate director of
legislative activism at the Electronic
Frontier Foundation we're a nonprofit
devoted to defending digital rights a
member supported um I particularly
manage all of our state policy at eff so
um very involved especially in the
California legislature both because
we're here and because California is
often a first mover um so really just
looking at um AI issues from the lens
that we always have right defending
users making sure that um that
independent people can break into the
break into the market and um just really
trying to make sure that uh legislation
doesn't pick winners and
losers I'm Joel uh I'm with the Milla
team um and so so Milla uh is up here in
part because we you know as Luther
alluded to are not just just a company
that builds products like the you know
Firefox browser that you all might be
familiar with but we're also you know
owned by a foundation and that's a
really crucial part of our work is you
know kind of fighting for an open
internet and especially you know things
like open source and as you know AI has
risen in prominence and missa has
increasingly you know kind of entered
that space um you know AI is taking up a
much larger portion of you know what
we're thinking about we had a Columbia
convening recently where we B where we
brought together 40 folks from you know
kind of the open source a community and
discussed many of the topics that have
kind of been highlighted here today so
um you know we we really feel strongly
about you know supporting the the open
source ecosystem and the open AI
ecosystem to making to make sure that
you know there is kind of Fair
competition for folks hey um I'm I'm gas
I am with Omar Network which is one of
the many philanthropic vehicles of Pier
Omar the founder of eBay so we long time
ago we come from from cagon Valley as
well and we just announced a new
strategy we're going to be focusing
mainly on Tech and the reason why I'm
really happy to be here working with
with Luther and with FR Mozilla and eff
is that um we think that what we want is
so you can you can succeed right we want
we we believe in competitive markets we
believe in like open source and tools
that um that that allow creators
entrepreneurs to to to succeed and we
like f rules of of the game uh the game
for everyone um and yeah so that's uh
that's that's what we stand for and
that's what we're happy to be here
amazing thank you so uh Joel can we
start with you tell us a little bit
about mozilla's new report on model evil
testing and open
source yeah so so out of this uh
Colombia convening that we did basically
you know as as I talked about we brought
together you know kind of 40 top minds
and you know the open source and AI
communities and I think you know if I
can underline one thing that that kind
of came out of this convening into two
reports and I'm not going to do it
justice you should go to the Milla blog
and check out the full report but I
think really it emphasized that when
we're thinking about Ai and I think
chair Khan talked about this a little
bit is that we can't just be thinking
about you know the the the standard
definition of Open Source doesn't you
know copy paste exactly when we're
thinking from traditional software to AI
so we have to think about kind of the
the full stack of AI as we're as we're
you know engaging with it everything
from you know kind of the the artifacts
to even like documentation and how you
know how open or closed that sort of
thing is uh for for users but also I'd
say the the the big take and this
probably won't be a surprise for many of
you is that um you know open source is
really really impactful and really
useful not just for spurring Innovation
but for doing things like you know
making it making it more possible for
you know researchers to find elements of
bias when you know someone is producing
when someone is producing something
making it easier for people to find if
there might be a cyber security
vulnerability and eliminating some of
the dependencies that you might get from
you know one single closed Source
company I mean I think many of us are
probably still thinking about the crowd
strike thing that happened a couple days
ago and I'm not saying you know open
source will necessarily solve you know
everything like that but it definitely
creates an environment where more people
can have more eyes on something and you
can help eliminate some of those risks
all of you sort of represent I and maybe
this isn't fair maybe some of you would
sort of brand yourselves as oh we're not
left or right we're sort of in the
Middle where we can we have issues where
we agree with folks that are right of
Center and left of center but I kind of
think of uh this sort of Civil Society
ecosystem system as as being a bit uh
left of center and I'm curious what your
sort of personal uh view of like P Doom
is like what's your probability that we
are going to face like existential uh
risk according to AI in in you know the
next handful of years or maybe the next
you know maybe give your own timeline
but um because one thing I've been
learning as I sort of get uh neck deep
in AI politics it's very strange to
predict which organization or individual
based on where they sit on the political
spectrum is going to fall here so I'd
love for you all to kind of tell us what
what's your personal P Doom how do you
think about uh these this tension
between safety and
openness um first um so humans are
really bad at predicting the future uh
and right um so I think that we need to
be careful and mindful like we shouldn't
not learn the wrong lessons from the
past right like we think of ptic policy
yes some things went wrong uh in the
last 20 years many things went wrong we
need to we need to address them but I
think to something that Lena said very
clearly L KH very clearly said for for
for a long period of time it's s that we
didn't have the rules to uh to make sure
that abuses were not correct so we were
not enforcing existing rules rules that
have been in the books for ages so I
think that we should not repeat that
mistake and and we should have guard
rails that allow creators and innovators
to create without constant fear of
liability I I believe that 99% of people
here are going to be trying to do the
right thing complying with all the rules
the best that you can so if we're going
to have a systems that that that that
create um that that create obligations
we should also treat good citizens
trying to do the right thing well giv
them the right to cure when they
discover that well it was a best effort
didn't work out and before get a
sanction let's try to correct it and and
and keep moving forward I think that's
the kind of like system that we need to
have uh that we need to have in place uh
and you know I think that in many of the
pie of legislation that we see around
the world that's very improvable um so
hopefully we're gonna we're gonna get
there the tldr of Gus refuses to answer
the question
next um yeah I'm not sure if you're
gonna let me off escate my way out of
this one but uh yeah yeah I I I think um
it probably frankly it probably depends
on how I slept last night is you know is
how I'm going to rate my P Doom today um
so so I think I I would I would say it's
definitely you know relatively low and I
think I think when we part of part of
the thing that um is hard with this is
when you're when you're thinking about a
single metric for something it's easy to
it's easy to look at you know all the
negativ so when we talk about P Doom
it's you know of course we're thinking
like Doom is a big word existential risk
is a big word but I also like you know I
was I was one of those you know I was
one of those people who kind of came up
reading Ray curtz while and you know the
singularity is near and you know I'm
definitely with describe myself as a
techno Optimist and I I think that um
you on that Brian Johnson
 does it look like it I got a
does it look like it I got a
little don't die baby don't die I think
Brian Johnson's coming to Y uh in a
month or two I I might might fly out
might fly out for that no so so I would
say I would say like look the there's
there's P Doom out there but like we
also got a factor in you pH hope or P
you know the world is going to be really
fantastic soon I mean this is my first
time back in SF in a long time and this
was my first day I got to ride in a
self-driving car and how cool is that
like I felt like it was in the future
this morning um and so like I have a lot
of Hope for technology and what it's
going to bring so I think we can't just
focus on the pom yeah I'll Echo what
they both said I mean and you said
personal so that's how I'm going to get
out of this right this is Haley and notf
necessarily speaking but I mean I think
you know for me in terms of priorities
for legislation like there are
definitely things that we could be
addressing right now with tools that are
happening right now and that's sort of
where I want to focus the priority and I
think if we set good principles for how
to address the problems that we're
seeing right now then you can kind of do
both at the same time right you can you
can think about the shortterm the medium
term and the long term so and I like the
idea of also articulating a more hopeful
future at the same time uh I'm gonna
stay with you Haley we were chatting a
couple weeks ago and I believe uh
Senator weiner is has arrived or he's
about to arrive uh we're excited to have
him here where's Scott weiner everybody
give it up for Scott weiner we'll see
him in a not we'll we'll see him in a
second we'll see him in a second but we
were chitchatting about uh you know sp
sp 1047 which just come up a couple
times uh today is the um you know eff
has sort of a a position of supporting
with caveats uh you know I think the the
community letter that sort of bubbled up
from from YC was sort of like oppose the
form is kind of annoying because it's
like they sort of hold a gun to your
head and they're like you either have to
you know you have to be for it or
against it and I I didn't know how to
I'm sure that there's sort of a version
that uh sort of the communities could be
happy with but I would wonder if you
could kind of unpack the the sort of if
you had a magic wand what what that
would look like uh the things that did
look good because I think you and I had
some agreement on some of that yeah so
well this is fun hello hello
Senator so yeah eff is in what we call a
supportive amended position as you say
the California legislature makes you
pick from like six very particular
positions um I think SP 1047 is a really
interesting bill it's a very Broad and
complex Bill and so um the position that
we've asked for is that um it creates
this Cal compute public uh cloud
computing cluster we really really like
that idea um and I think uh we just
think that it deserves to be the focus
of the bill like that it's worth getting
that right um and then it's also worth
having the other conversations that are
in the Bill about how do we you know how
do we address safety um and those we
would like to see those condition those
conversations sort of continue with more
stakeholder input we'd love to have you
know folks from um from this community
engage in that um you know there were a
lot of AI bills in the legislature this
year um and I think just more
conversations needed right if you even
think about the number of definitions of
AI that we saw come through in the
different bills this year I think that
entire regulatory conversation is very
important we want to get it right we
want to be sure that we're thinking
about how to address it now and how to
address it in the future so um that's we
would like to see the bill focus on the
public cloud computing cluster okay I
said we're going to speedrun this panel
to get us back on track any super quick
questions from the
audience gotta be
quick uh hi I'm Brandon I'm one of the
co-founders of square. um in specific
regards to SB 1047 um I'm interested in
all of your takes or one depending on
speed uh on the reach of the bill and
it's ability to impact an end engineer
based on things in the bill and
essentially and I believe as we all
discussed openness and how much open
markets competition are important when
there is that type of I'll say liability
to the end user often people will never
even enter the market so they're like we
don't are we not able to observe the
void that exists if something like that
came in yeah I I I I just mentioned you
know trying to trying to understand the
the the counterfactual you know like who
doesn't go into you know developing
something is is super hard to calculate
so I mean it's certainly you know it's
certainly possible um I I think that's
you know I'll let I'll let the senator
talk about you know the potential impact
and kind of pwn this off on him so like
look the I I think I think what I would
emphasize is that like we're here in
Silicon Valley you know a lot of amazing
things get evented here so of course
California legislation is going to
matter a lot I I don't think that's the
intention of the legislation but if
we're in a context where like we're
debating on what can to be the effects
and we've seen that the Jud is making
interpretations of legislation that are
often contrary to what's clearly the
intention of the legislator um we need
to be very careful in how we Draft rules
to make sure that that doesn't happen
both that the intention is clear and
that there's no possible argument of
like what is intended and if like the
fact that just the fact that that's a
question that's the case that to me
invites a reflection on how can we
improve things so so we know that that's
not what's intended
and that is our panel thank you very
much thank you all for
participating and um if I can welcome to
the stage uh Jake heler from case Tex
well Jake is uh getting set up I wanted
to say Jake has one of the coolest
stories that uh we went into detail over
on my YouTube channel and uh Jake is
also one of the people who got the
earliest access to gp4 and one of the
things that was most remarkable about
that was uh I think you said once you
got it you realized oh hallucinations
came down to a level where it almost
never happened and we could actually use
it and he pivoted the company into a
tremendous uh more than $600 million
exit uh so you this is one of the first
and most powerful examples of practical
AI being used day-to-day uh in one of
the biggest Fields out there so Jake
take it away um so hey everybody I'm
Jake uh I was a Founder in the summer 13
batch which is forever ago and um I was
a lawyer before I started this company
uh and what we decided to build
especially with this new technology is a
legal AI assistant the idea and this is
not that different from probably a lot
of the applications you're building for
your various Fields is uh for there to
be almost like a new employee at every
Law Firm government agency nonprofit um
and corporate legal office that is uh a
new like legal employee that you can
assign tasks to and like a fellow
employee it'll be able to do that that
that task I'd say right now like a
junior associate but um unlike your
human colleagues it is able to do it at
a speed that is uh super human so we're
GNA go through a few examples um that
I've kind of teed up it's kind of for
fun we're going to imagine that we are
representing Elon Musk as part of his
litigation against Twitter and uh you're
trying to is you're trying to pull out
of the merger agreement so the first
thing I do is ask you know can you pull
out of a merger agreement when the
seller made material repes
misrepresentations under Delaware law
that was his that was his argument and
we'll get our AI kind of researching
that what will happen behind the scenes
is that it is um conducting dozens of
search questions taking that that
question doing dozens of searches
against a legal database pulling back
hundreds of answers reading each of them
in full all the cases rules regulations
and statutes combining all the
information gained from that activity
and turning that into a memorandum and
while it's doing that maybe we'll ask
some other questions uh for example
during law uh legal proceedings uh you
may at times gather evidence what's
called Discovery and so here we have um
elon's texts regarding uh the
acquisition and we might ask for example
what are all the features Elon wants to
add to X SL Twitter and just for fun I
actually picked this off of a New Yorker
article these are not as real texts as
you'll see soon
um
uh and uh so they're funny and they're
not real but they're still you know
still really great um and at the same
time as I'm doing that I can upload say
the complaint from the
case and ask it to make a timeline of
all of the events described in the
complaint related to crypto spam
and and so you're see able to do here
and I only have like five minutes is
it's now completed the kind of legal
research task that would have taken me a
good part of a day and here it's kind of
written out its answer um citing to the
relevant case authorities to which I can
from the relevant jurisdiction I can
open up and research and read myself um
it has gone through the text messages
and is able to not just answer the
question but also so uh one are the kind
of funnier ones delivery tracking EG
Pizza it's able to pull out the exact
part of it and highlight that for me to
confirm you know not that just kind of
making stuff up but that it's also kind
of finding the right
answers um it's pulling together this
timeline for me um each of these tasks
take lawyers typically the better part
of a day um and to move off the Elon
example you can also do some things that
are a little more powerful with it for
example um I've uploaded here my series
a document that I got from USV back in
2014 and have asked to compare the
series a term sheet against um all the
different um actual documents that we
ended up getting to the degree which
they deferred because sometimes there is
drift between what was agreed to in the
term sheet and what was agreed what
actually happened in the actual you know
final documents and sometimes it's was
actually kind of hard to get but um you
know so a lawyer May sit down and you
can see here it's kind of listed out
everything the price per share the
closing date the dividends liquidation
preferences you can kind of analyze
through this table which are the
different things that it ended up
catching are potential differences
between the uploaded term sheet again
with the ability to say um you know go
into the right of first reviews all and
caught here that the closing date was
intended to be January 16th 20 whatever
it was uh 2015 but actually ended up
being January 20th and that was a
difference between the term sheet and
the so um you're able to do these kinds
of things that that aren't just
difficult for lawyers who do but in some
ways impossible um it's hard for a
lawyer to keep everything in their head
about the term sheet while also reading
through every one of the many many pages
of documents but for an AI say an AI
with an incredibly long context window
um of millions of tokens you're capable
of loading up all of the documents
having them compare against each other
and make those kinds of comparisons um
we as Gary said we exited to a company
called Thompson Reuters which is why it
doesn't say casex up here but has this
kind of funky orange logo um we now have
over 30,000 lawyers use us daily um and
uh this is one one of the early examples
I'd say what I expect to see across the
industry and a lot of things you guys
are building of this assistant
infrastructure assistant kind of agentic
approach uh taking place within a
profession and within professional
workspaces to get real work done so
thank you
[Applause]
last but certainly not least it's sort
of the one we've all been waiting for I
think is uh Senator Scott weiner um and
we have uh interviewing him on stage is
sharen gafari from Bloomberg I'm going
to read their bios and invite them up to
the stage so uh Senator Scott weiner is
the California state Senator
representing District 11 that's San
Francisco um it also includes parts of
San Mato County uh he was elected in
2016 quickly established himself as a
prominent voice on issues critical to
California's future particularly those
relevant to the tech and startup
Community um as chair of the Senate
housing committee weiner has been at the
Forefront of addressing California's
housing crisis he authored land Landmark
legislation to streamline housing
approvals including sp35 and s423 which
have resulted in thousands of new homes
across the state
[Applause]
yay prior to his Senate tenure uh wer
served as a San Francisco Board of
Supervisors and work as Deputy City
attorney his background also includes
experience as a litigation attorney in
the private sector bringing a unique
blend of private and public sector
perspective to his role uh he's known
for Innovative approach to policymaking
tackling issues from Transportation
climate change mental health addiction
treatment recently turned his attention
to emerging Technologies demonstrating
his commitment to addressing the
challenges and opportunities facing
California's tech industry with his
Harvard law degree and focused on
Forward Thinking policiy Senator weiner
brings a perspective that should
resonate with our community and shine
gafari is a reporter of Bloomberg News
specializing in artificial intelligence
ethics and policies she joined Bloomberg
in September 2023 after six years at Fox
where she was a senior correspondent
covering social media industry and major
tech companies we will have uh some time
we we're GNA have it's going to be a
little bit like the uh Lena and Jonathan
panels sh will interview um uh Senator
Weiner and then we'll have hopefully
plenty of time for for Q&A and we'll
we'll probably move to to end about 155
and a special uh surprise guest speaker
at
[Applause]
155 all right um so thank you for doing
this talk with me today Senator weiner
um your bill is just a little bit
controversial in Silicon Valley uh for
those who don't know but I imagine many
of you do it's a bill that would require
any AI company De deploying a certain
threshold of compute power to test
whether its models would lead to
catastrophic risks among many other
Provisions but that's sort of the main
gist of it um I want to start off just
by asking you why now so AI especially
generative AI it's still very nent right
um why did you feel a sense of urgency
to do this uh sure first of all thank
you for having me uh uh today uh really
appreciate it and I also just want to
say thank you for um all of the
engagement by a lot of different people
in a lot of different ways on this bill
it's been uh it's been uh it's been
interesting and it's uh it's a little
different I'm used to like housing
policy where you have um a lot of just
um a different kind of engagement at
times um sometimes less a little less
constructive um and so I I really and I
appreciate also to the folks who are
critics of the bill we've really um
gotten a lot of constructive uh um
feedback from from folks who are not
fans of the bill and and we really
appreciate that um as well in terms of
why now as human beings we have a
tendency to um not deal with uh
potential issues or problems or
challenges until uh something uh bad
happens um it's just I think sometimes
how we're wired as uh human beings and
we've seen this in the technology uh
space around data privacy around social
social media about any number of issues
where we really did essentially nothing
and then had to play uh catchup
afterwards uh and by the way Congress
still has not passed a federal data
Privacy Law in 2024 um we had to do we
did that in California in 2018 and it
was controversial and a lot of folks in
the tech world were opposed to that bill
and told us that everyone was going to
leave California if we passed it we
passed it that did not happen um and we
set a standard that I think was a really
uh powerful one um I also um so I think
it's about trying to understand and get
ahead of safety risks uh and doing so in
the context of the reality that AI is to
me incredibly exciting uh what's
happening now is incredibly exciting the
prospect of being able to solve problems
that we have been unable to solve as a
society that of AI contributing to those
Solutions is incredibly exciting I
fundamentally believe that AI can and
will make the world a better place but
with any incredibly powerful
transformative
technology there are always going to be
risks it's not unique to AI there always
going to be risks and so we should try
to understand those risks people who are
building incredibly powerful models
should try to understand those risks as
they have already k committed to do all
the big Labs have committed to do this
testing uh and to mitigate if it's
possible not to eliminate risk it's not
we can't eliminate risk that's not like
the world the reality but trying to see
if there are ways that we can reduce
risk so really accentuate the good um
while understanding uh the potential
harms and the and why now in California
because I don't have any confidence that
the federal government is going to do
this like I said Congress has not passed
the data Privacy Law yet um President
Biden issued this executive uh order um
which uh the other person running I
won't he who shall remain on names um
has already said or the Republican
platform has already said that they're
going to get rid of that executive order
the Republicans have been trying to get
rid of nist uh and so we have a
responsibility in California to
lead so I know that you've done some
polling or there's been some polling and
there was you know there there is some
support people say you know in the
general kind of public for this bill I
know it's pasted California house by
wide margins but there has been a lot of
push back in the Tech Community a lot of
concern that this could stifle the
startup ecosystem particularly open-
Source AI so I just want to ask did you
expect and anticipate this level of push
back on the bill from the tech community
and what do you think are the most Fair
criticisms of the bill and what do you
think is maybe you know the biggest Mis
perception of of it um yeah so the the
polling has been off the charts
including among Tech workers um but
ultimately as I've said um many times
this is this isn't about to me it's not
about pulling it's not about just
winning winning the fight I want to get
this right that is my goal to get this
right and that's why from the very
beginning um I've tried to send very
very powerful signals that we have an
open door and that we want constructive
feedback about the bill because we want
to get it right and I am so deep
grateful to the folks who have really
worked hard to provide that feedback
including folks in the open source
community and we've made a series of
amendments uh in direct response to some
of that critical constructive feedback I
no no I I knew that there people were
going to have opinions about this bill I
was I you know the bill arose despite
some of the conspiracy theories um
online and there have been a lot of
conspiracy theories about this bill this
bill arose uh organically just
conversations I was having with folks in
the AI space Frontline technologists
startup Founders um academics um you say
which ones there was I I attended a
series of salons dinners there were
actually a large number of people and
that's not to say they all had the same
opinions um but that that's sort of how
this uh came up uh and we then started
Haring a lot of the the labs and DCS and
others and academics say hey can you
tell us what you think tell us what you
think so um it was not the other way
around no one came to us and said here's
a bill with a ribbon wrapped around it
please introduce it we came up with it
and then we started soliciting um uh
feedback uh on it um and so and I
actually took the um pretty like
extraordinary very rare step I've
actually never seen it done before where
um in last year's legislative session
2023 on the last day of session I
introduced what's called an intent Bill
basically a an outline of the bill we
like blasted that out to the world and
and actually put it into an official
bill um not not because an outline you
can't pass an outline but we wanted to
literally like just put it be super
transparent about what we were thinking
about to try to generate um that
feedback and critiques um of the bill so
we've tried to do that uh from the very
um uh very beginning um so no I'm not
surprised that there's push back I what
what is I don't want to say surprised
because I've been around politics long
enough to never really be surprised um
by by any kind of politicking there's
been a lot of great um analysis critique
feedback about the bill there's also
been a lot of inaccurate information
just to be blunt especially on Twitter
um shockingly um but elsewhere as well
and um what specifically do you think is
the inaccurate information that has been
spread about this um there have been we
I was very intentional when we crafted
this bill I rejected the idea that that
you would have to get a license from the
government in order to train or release
a large model there are people who think
that you should have to get actual
permission from the government I
rejected that um out of hand and yet
there are still people who are either
stating or implying or suggesting um
that you're going to have to get
permission from the government in order
to train or release a large model um
there um have been uh statements that uh
model developers are going to go to
prison um if their if their model causes
some sort of harm which is false it is
absolutely false the only way that this
bill interacts with the criminal law is
if you commit perjury meaning that you
intentionally lie to the government just
like if you lie on a driver's license
application or a public benefits
application or on your taxes that's a
crime you're not allowed to like you
know with knowledge intentionally lie to
the government um and and yet so you
know there there were statements about
developers going to prison because their
model somehow contributed some sort of
harm that's not true um I think that
there have been significant
exaggerations about the scale of the LI
potential liability uh in this law for
for model developers and first of all it
only applies to you have to spend at
least 100 million which is tied to
inflation on the training uh we're
talking about the you know large Labs um
but in terms of liability um there are
two ways where I think it's been
inaccurate The Narrative about liability
um first of all um and I will say this
until I'm blue in the face even if some
people are not always hearing me um
model developers can be sued
today if your model if you release a
model whether it's llama or whether it
is a much smaller model that wouldn't be
covered by SP 1047 if you release a
model model someone uses that model in a
way that you know burns down someone's
house or shuts down the Grid or this
something that is that causes harm you
as a developer can be sued today not
just in California but probably in all
50 states under just general tort law
now with AI models that's that'll get
litigated in the courts and the and it's
going to be it's a new developing area
of the law and the courts will set
parameters um in terms of under what
circumstances an AI uh developer can be
a model developer can be um held liable
uh but but in terms of the risk of
litigation that exists today and it is
dramatically broader risk than what's in
our bill because what's in our bill only
one human being uh no AIS and only one
human being can file lawsuit and that's
the Attorney General of California
there's no private right of action for
you know a lawyer to run into court to
file a lawsuit to only the attorney
general and only in extremely limited
circumstances and so that's been another
area where I think that the the risk of
litigation has been significantly
exaggerated under this bill while
ignoring the current tort law liability
uh that is a risk litigation risk for
all AI model developers today I think
the most common and we're going to leave
some time for questions shortly the most
common concern I hear from startup
Founders and developers is that this
could have a chilling effect on open
source that um that requirement that you
mentioned that's not far away from
something like a llama 4 might be right
meta's open source model and that many
developers rely on um these bigger more
well-funded companies to release open
source models so if they're discouraged
um to release those models in the future
that would that would affect startups
what do you say to that criticism yeah
I'm very sensitive to that I I want to
be just I'll say I've said it before
I'll say it again I support open source
I I think open source brings so many uh
benefits uh to The Innovation uh
ecosystem and I'm not her goal here is
not to shut down uh open source um we
have some Provisions in the bill and
we've made amendments based on open
source Community feedback uh for example
we clarified very explicitly that if if
the model is no longer in your
possession you are not responsible for
ensuring that it can be shut down that
was a really significant concern that a
lot of of Open Source folks were
expressing to us that once I open source
the model how am I responsible to be
able to shut it down that was never Our
intention uh but we realized that we
could say that more clearly in the bill
and so we did that we also were much
clearer about what counts as a
derivative um model and so if a model is
changed Beyond a certain point it's no
longer the responsibility of the
original um uh developer um and so
uh you know we've tried um you know in I
think in good faith uh to try to address
the concerns uh of folks in the open
source uh space uh and I you know I I I
I don't think that this is going to shut
down open source Innovation for and also
for some of the reasons that I said you
know before and and also to be really
clear all of the large Labs including
meta have committed to do doing safety
testing you can read out it's all in
writing what they said they what they
committed to in Soul South Korea two
months ago um where in addition to the
White House where they committed to
doing the safety testing they committed
that they would not release models if
there was going to be some huge harm
that they you know that that that was
not able to be mitigated it's all in W
and they've committed to do this
including
meta and so that that's that's why to me
uh if they're already we're our goal is
to is to require them to do what they
have said they're going to do um I also
said uh that if if what we're requiring
is a mismatch doesn't exactly pair up
with the public commitments made just
tell us how you think it's different
we're happy to listen to that so the
bill is still open to amendments is my
understanding and actually just today um
the first kind of major AI lab that I've
seen came out to support the bill with
the caveat that they want some changes
right and that's anthropic so do you
want to talk a little bit about that and
are you open to making the changes that
anthropic has asked for yeah so
anthropic last night sent in a letter
saying it was talking uh as the uh as
eff told us before they checked the the
box of support if amended meaning we
will support it if you make the
following amendments and there were
about three pages of detailed amendments
and uh and I want to say I'm very appre
itive that anthropic and they're not the
only one but the anthropic uh took
seriously when we said give us feedback
tell us what you want to see change in
the bill and anthropic came with some
very detailed um uh amendments uh and
this is you know anthropic also you know
a a lab that is prioritizes safety and
has the I think the largest uh the
highest the most capable model around
today maybe I'm wrong about that for now
but that changes every yeah I'm sure it
changes every week um and so uh and we
are um I'm very open to their amendments
they have was some very very thoughtful
and interesting amendments and I'm uh
I'm I'm generally positive about their
amendments and you know one of them is
eliminating the frontier model division
the fmd which is cause um some
consternation and I'm I'm open to doing
that we have to because of the late
stage of the bill it's gone through a
bunch of committees with amendments from
committees and so we have to for some
kinds of amendments get sign off from
the committee that passed the bill so
it's not exclusively I'm I'm not the I'm
not the the the reigning monarch of the
bill exclusively it's like a group
effort at this point um but we are I'm
very open to the Amendments that
anthropic has proposed and there's still
time if people have you know and some
people there are some people who uh um
who I I won't name names but some people
who have made some very concrete uh
suggestions and we really appreciate
that and there's still time but time is
running out so time is find him after
this talk do Q&A thank you Shir we gotow
in time so I want to move to to Q&A how
many how many uh Founders are here raise
your hand if you're a Founder raise your
hand if uh put your hand down if you're
not using AI in your
company put your hand down if you don't
live in San
Francisco okay so a lot of lot of
constituent AI Founders in this audience
so I'm eager to hear your questions uh
let's see some
ants thank you so much for being here
and uh I think your openness to feedback
is commendable and I think we all
appreciate it um I don't think a lot of
us here are going to train models from
scratch to hit that threshold so let's
talk about startup Founders there's a
higher probability we're going to
fine-tune
all right so from what I understand and
I read all of the text this morning
summarized it did a bunch of little
analysis even if you're fine-tuning
you're considered a model
developer and if you're fine-tuning
you're still subject to a lot of
compliance now we did some rough math on
the compliance that's like half a
million a year roughly in all of the
compliance work necessary in order to
find tune and release a model that is
derived from a much bigger model that's
actually so it's I don't want if I could
just I can wait till you're done go
ahead no the in terms of fine tuning it
is those responsibilities would would
kick in only with a quite significant
amount of fine tuning U so there's a
threshold in the bill and and that
doesn't have to stay that way in in the
bill right that was again that that's
the kind of feedback we won't I think
it's very
unlikely that it would be dropped like
that um but that's certainly a
appropriate topic of conversation but
the but it is a a significant amount a
fine tuning you'd have to do for it to
become effectively a new model it's not
about you know modest amounts of fine
tuning I've got it yeah that would be
helpful to clarify very clearly I think
it's very clear that threshold in the
bill um I I think as was just pointed
out that there was some concern that the
um the frontier model division could
make an adjustment to that over time it
could also raise it over time and this
is always by the way um with any kind of
Regulation the push and pull is always
that if you make everything set in stone
then people will I think legitimately
say wait you're setting it in stone what
if the world looks different in 3 years
and then if you say okay we're going to
give some flexibility so that you can
look at changes in the world then people
say but now we don't know with certainty
what it's going to be uh in a few years
so that's always a balance that you try
to strike to have not too much
flexibility but at least some
flexibility I got a question back here
uh hi I'm Evan CEO of edge um first of
all I do want to say thank you so much
for engaging I'm Okay naming names I
actually been engaging with your office
and sending over specific suggestions I
also put together a letter that hundreds
of YC Founders um signed so I I can
vouch for you really are engaging and
thank you for that um so my main fear
with this bill is that will become squa
for software you're one of the leading
proponents of squa Reform and so you
know that squa was not originally
supposed to be squa the friends of
Mammoth court case turned it into squa
and now even when you're not violating
squa the cost of producing thousands of
pages and the risk of being violating it
is you know is a huge impediment to a
lot of things like housing High-Speed
Rail and again this came from Court made
law not from legislative made law and so
there are two main areas where I'm
curious if you're open to some changes
here the first one is the definition of
AI the definition of a model in the bill
is a is a piece of software that can
implicitly based on inputs and outputs
Define goals that can plausibly Define
almost any software and so I'm wondering
if you're open to changes to the
definition to narrow it very tightly to
apply to a much narrower range of
Technologies the second one is that
right now it's really focused on
liability your team has said that
they're really focused on connecting
this to negligence law I think that's
great I'm wondering if you're open to
instead of substantial you know
reasonable likelihoods of harm or
something connecting it to proximate
cause which is the negligence term and
connecting it to a liability Shield
where when you release testing you're
sort of making this trade where you're
deciding how much testing you want to do
you're deciding how much cost you want
to put in to testing and in exchange as
long as you comply with the O the
testing of your standard you get a
liability Shield that protects you
within that scope and I think those
three changes would actually turn this
from the potential SQ software to
actually a positive for the industry
that provides a lot of clarification and
makes it easier to do business as an AI
company in California um okay so a few
things hopefully I'll remember
everything um yes you can remind me if I
forget um in terms of the definition of
AI we actually um we had one a
definition when we first introduced the
bill and then through the legislative
process because there are um as was
noted a number of AI bills large and
small um there was a uni the legislature
the two houses working together came up
with with a uniform definition of AI
because what would be bad is if we had
like 10 different definitions of AI in
different bills and so it was made
uniform and I think it was looking to
other sources of law so um you we're
always willing to you know look at um
specific uh ideas one thing I will note
on that is because uh the only applies
to models where at least $100 million
tied to inflation uh was spent on
training that excludes an awful lot of
uh potential uh potentially included
models so in ter we are we are o open to
looking at you know we making sure that
it conform or that it is as you know
close or more conformed with negligence
law that's always been and so that those
are that's like a an ongoing topic of
conversation in terms of a liability
Shield um it does shield you from
liability under this law terms of if
you're talking about a broader liability
Shield to like eliminate like existing
liability we're that's the bill does not
do that we're not changing other
potential liability that someone uh
might already uh uh face and then in
terms of squa yes um first I don't want
to assume that everyone knows what squa
is squa is the California Environmental
Quality act it was a law that was passed
signed by Ronald Reagan a republican in
the early
1970s um with which uh basically says if
you're going to do something like for
example if you're going to build a dam
or a new freeway you have to evaluate
the environmental impacts to inform your
decision about whether to do that over
the years I've described squa as the law
that swallowed California uh because now
if you are trying to change uh the if
you want to rebuild the rotted out front
porch in your home uh that can be
potentially subject to sequit and it and
it has undermined um climate action
building new housing public
transportation Etc um I would actually
disagree that this will be I don't think
this is going to be the in any way the
the sqa of AI um this is so well defined
in addition this bill does what many SQL
reform Advocates have been advocating
the the biggest problem with sqa is that
anyone can file a lawsuit even if you
are you don't even have to disclose Who
You Are you can have like a fake email
account and file a sqa appeal and a sqa
lawsuit it's it's absolutely absurd and
a lot of proponents there are proponents
of SQL reform who have said let's just
because SQL does some important things
let's just say that only the Attorney
General can file a SQL lawsuit that is a
a pretty significant that would be a
huge piece of SQL reform if you couldn't
have private lawsuits under SQL that
only the Attorney General could enforce
it and that's what we do in this bill so
I uh I I dispute that this would be the
equivalent of squa for AI I think uh
this is profoundly more narrow than what
SE let's go to another
question I got uh Jeremy Nixon founder
and CEO of omniscience and yeah first I
wanted to thank you for the hundred
million ceiling that was a big
Improvement to floor second floor yeah
floor um and I think it s a lot of
people's very deep worry that open
source would not be releasable or that
it would be hard to you know work
through government regulation on mass um
my question is about meta's decision not
to release their multimodal models in
the European Union I have a sense that
there's a potential world where you know
models like llama 405b are launched in
the United States perhaps parts of
Europe that are not you know in onerous
regulation but not in California and I
guess um one of the reasons for that you
know is related to privacy but I guess
um would you be working with you know
Yan Lon and folit meta to ensure that
Yan laon for example the head of
Facebook sayi research or and folk at
meta to ensure that s SP 1047 yeah isn't
onerous to a point where they would not
release their models in California to
begin with yeah because all of us are
are fine-tuning these models and it's
essential to our businesses and and I'm
sensitive to that and I know you know a
lot of people are relying on uh on on
meta and and making sure they have
access this bill is so much more narrow
than the European Union AI law um and as
you noted there are other issues there
as well it's not just the AI law their
their data Privacy Law is a lot Way
Beyond what California um requires so um
I don't I don't put them in the same
category um and we've I think our bill
has always been fairly focused and
narrow and I think we made it more so
over time um I also um you know
California is uh uh so one of the um
things we've heard is companies are
going to leave California startups are
going to leave California they're going
to develop they're going to train their
models release their models in you know
whatever in Texas or Florida or some
other state that doesn't let women
control their bodies um and that Outlaws
trans people so everyone's going to go
to those states that that's what we hear
um I don't I don't know that I agree
with that um but in addition the bill
does not what triggers the Bill's
applicability is not where you train
your model or where you release it it's
are you doing business in California and
so you can go to Austin or Miami or
Omaha and train the model there and do
everything there unless you're going to
have no connection to
California then you're covered I'm
skeptical that that would happen but
fundamentally this bill I don't think
you can compare this bill to the AI act
or whatever its official name is in the
EU this is much much narrower than that
let's try to do two two quick quick ones
uh you got the mic uh Hey Senator as you
know I'm a huge fan thank you again um
your your work in affordable housing has
has helped more working-class California
than pretty much anybody else I'm more
of um uh thank you for engaging so much
with 1047 uh as you know a concern I
have is that so many crucial tools are
capable of both utility and harm uh as
an example I was talking to the head of
DARPA bio yesterday and he said wow you
know climate change is coming um we need
to be able to re-engineer crops for
better drought resistance pest
resistance all of the AI tools that
would really unlock this probably could
be used for for critical harm and he
said I would be dismayed if researchers
large and small Across the Nation did
not have access to these tools which
which inherently could do harm uh so I I
I love so many of the controls youve put
in but I am curious if fundamentally if
there is a a a a a duty of reasonable
care which is tied to the quality of the
Safety and Security protocol uh or or
the responsibility doing no critical
harm how do we navigate that with tools
that should be available to the general
public but which could fundamentally do
harm yeah and and I thank you for your
uh very detailed engagement I appreciate
it um uh so there's been an uh some
people have and I understand we all use
sort of um jargon and shortcut language
um and we've seen this in some of the
letters from some of the big tech
companies about how the um the bill
requires that you like certify that it
can't do anything harmful or whatever
similar language to that which is not
what the bill uh requires right now we
have a reasonable Assurance we're
looking at that um language and I think
and I I I would not be shocked if that
um got uh massage again we're not
looking to the goal here is not to
eliminate
risk like you know you eliminate Risk by
living in your basement and never go and
even then there's risk because the
ceiling could still collapse on you or
building could catch on fire so there
like literally there's no human way to
eliminate risk life is about risk but we
also as human beings like to try to do
things if it if it's if it's reasonable
if we can find a reasonable way to make
the risk less right like lower the risk
of the builder of the building Burning
Down by and or the roof collapsing on
your head by having building codes we
don't put building codes in place that
are like over the top uh to to make it
because that might make it impossible to
build the building and that's our
Approach at 1047 we want to just you
know create incentives to take
reasonable steps to reduce risk not to
like you know you have to sign an oath
that nothing harmful can ever happen
from your model that's not what we're
asking for uh and again if there is
specific language that just like with
the um the idea that uh if you release
the model and you no longer have
possession of it um you know whether you
should be responsible for being able to
shut it down we our intent from the
beginning of the bill was that you
shouldn't be responsible if if you no
longer have possession but there were
people who said the language isn't so
clear and so we went and we made it
clear and that applies to other um areas
as well because there are times when we
think it's clear but someone else says
you know what this is a little vague um
and we want to know that and so D I
appreciate like that you've been you
know trying to help us with that I think
we're GNA do one final question we're
gonna have be really quick because we're
about 10 minutes over and I want
everybody to stay in their seats for the
uh uh surprise speaker at the end uh I
want to know who the surprise
speaker hi Senator uh Angela Hoover from
Andy
uh I really appreciate you being here
and being open to having the
conversation I think my biggest concern
with SB 1047 is to do with competition
and I think compliance got brought up
before it's a huge concern for me I mean
you know monopolies if you take search
for example already have so much going
for them and when it comes to small
startups the cost of you know half a
million dollars a year is just not
feasible for little Tech and so I'm kind
of curious to get your opinion on if you
think that this is going to stifle
competition and actually lock little
Tech out of being able to compete in the
AI space that that is certainly not my
goal and I don't think that it will and
that's again why we we put this the
hundred million training threshold if
you are spending less than $100 million
and it's tied to inflation if you are
spending less than $100 million to train
your model then you you're not you're
simply not covered by the bill um at all
um there are threats uh to competition
um and and there are and we know that
there's a history in Tech of big Tech
big tech companies trying to squash the
little tech companies and those threats
are right now with vertical integration
and and around uh chips and and compute
and Staffing and data and and the big
labs and the big companies gobbling up
um a lot of those resources that to me
is the the threat not not this bill and
I will also um stress that Google and
meta so not not just not just meta but
Google as well are opposing the bill so
the companies that have a history no
offense it's just like what it large
companies what they do and that's why we
like to have Pro competition laws uh
which I appreciate that YC has been a
strong supporter of pro competition laws
we have to try to and and antitrust and
amazing people like Lena Khan to try to
stop the big companies from engaging in
that predatory behavior this bill is not
going to empower uh that behavior and
that's why we limited it to these really
huge models we think that startups um
are are overwhelmingly not even going to
be covered uh by the law at all all
right everybody please uh give a great
thank you to Senator weer thank you so
much for coming CER appreciate it anine
gabari from
Bloomberg okay so uh we were uh you know
in in the chaos of lunch I was sort of
looking at the crowd and I hope hope
you're able to stick around cater and fi
some of these questions um I was looking
around the crowd and recognized an AI
celebrity and I asked if he wouldn't
mind instead of me doing the closing
remarks you guys have heard from me
enough today uh would just uh give us
some closing thoughts uh Andrew is the
is a globally recognized leader in
artificial intelligence machine learning
he's the founder of deep learning AI
Landing ai ai fund as well as co-founder
and chairman of corser uh Dr previously
LED AI teams at Google and BYU where he
was instrumental in developing
transformative Technologies like Google
brain he's an ed professor at Stanford
uh he's educated Millions through his
online courses democratizing access to
AI knowledge his pioneering work spans
deep learning computer vision robotics
AI applications across various
Industries um um with 200 published
research papers numerous accolades
including being named times 100 most
influential people um Dr continues to
shape the future of artificial
intelligence as both an innovator
investor uh his mission is to build AI
systems that enhance human capabilities
and improve lives on a global St we're
so honored to have you make an impromptu
closing out of our amazing day today
Andrew please thanks
thanks Luther for inviting me up and
Gary Luther for um having me here so I
was actually taking some notes as the
senator was speaking and I want to share
some thoughts on AI regulation and also
address some specific points that
Senator wner had weer had uh mentioned
so as some of you know I've been
dismayed at the assaults on open source
software they been seeing um around the
world so over the last year I've
traveled to many nations capitals
including Washington DC to speak
Regulators about Ai and AI safety um to
be clear I'm actually Pro AI regulation
and pro AI safety for example the uh The
centor Bash the Washington DC but I
think that um the Senate just passed the
Defiance act which I think is a very
positive step to enhance AI safety by
making it more difficult uh you know to
to try to squash out non-consensual
non-consensual de fake porn so I think
that's a great move by by our Nations
Capital just just pause the sence um and
when I look at different Nations I think
the biggest mistake that many
legislators are making is to regulate um
technology and applications in the same
way and it's important to distinguish
between technology and applications so
one of the tricky things about AI is a
general purpose technology I think
there's a question on this just now as
well meaning that AI is useful for a lot
of different applications and so that
way we regulate for AI safety must be
different than way than the way we
regulate for narrow use Technologies to
make an analogy um an electric motor is
a technology with tons of applications
when we take the electric motor and you
put it in an electrical vehicle or a um
blender or a dialysis machine or to
steer a smart bomb those are the
applications and I think we can look at
applications like blenders or electrical
vehicles and say how do we want to make
sure that your blender is safe or your
EVS sa and what outcomes do you want or
often not want if you know for people
are trying to build smart bombs I would
be good regulations but if were to go to
electric motor maker and say I need you
to make sure your electric motor is safe
electric motors have no way to do that
and the only response would be to say
well I don't want to make electric
motors anymore if if I can't stop
someone from taking my electric motor
and adapting it to harmful application
AI is like that too when someone trains
a large language model or a large
multimodal model or other generative
model that's a technology
when someone puts that into medical
device or into social media feed um or
into chatbot or uses that to generate
political defects or non-consensual
defect porn those are applications and
the risk of the risk of AI is not a
function it doesn't deter it doesn't
depend on the technology it depends on
the application this is why I think
Washington DC did such a nice job
thinking through the defin act because
that's a you know really problematic
application of AI so let's use
regulations to squash that and just as
you can't ask electric motor maker to
guarantee that the AI will never ever be
used for harmful purpose I think we as a
large language model provider to make
their model safe from a scientific
technical point of view I actually have
no idea how to do that um and and in
fact in the manner specifically of
sb104 in the manner of
[Applause]
sb1 um in Man of sb147 I've actually
read the statute quite carefully myself
and there are causes in it that as a AI
developer myself I honestly have no idea
I honestly don't know what I'm supposed
to do to to comply with those and again
just to be clear there is a difference
between general purpose technology and
narrow use Technologies so for example
C4 explosives has a relatively narrow
set of use cases and so if you regulate
the technology it does slow down all of
those use cases of C4 but that might be
a good thing um since we're in San
Francisco where unfortunately we really
struggle with the fentol prices I want
to say fentol chemical precursors have a
relatively small number of use cases you
know you can't do much more than MC
fentol and a few other things with them
so if you crush innovation in the
manufacture of fentol precursors which I
also support you will slow down the rise
of fentol it seems like a good thing in
many places but if you crush AI
Innovation you slow down tons of
beneficial applications and in fact I
think that the ass and open source
represented by many regulations around
the world will make AI less safe because
I've been speaking frequently with
academics from UC Berkeley from Stanford
from many of the Great universities in
California and around the world one of
the key ways to keep AI safe is to keep
it open because academics need to study
these models in fact s mentioned social
media a lot of problems of social media
we collectively discovered only because
a smart reporter or a smart academic
really dug deep and found those problems
so one of the best ways to make AI less
safe if that's the go is to shut down
open source and shut down open AI which
is why I'm very concerned that the
assaults and open source in the false
name of safety will actually make AI
more
dangerous now to to add just Fe the S
remarks um I would I I I appreciate your
remarks and I appreciate you coming here
to speak um I respectfully submit that
um your mentioning the hundred million
threshold is not the full story it is
true that my I'm not a lawyer not giv
leg rice there is hundred million dollar
fesh H holds and I am concerned like
many in the entrepreneur Community are
that if the unelected fmd changes that
threshold that then people without
spending $100 million dollar will become
subject to liability is Chris Langer
here oh stand up so Chris actually um
has one of the best analyses of uh sb147
did I encourage you to to to find the
line and feed what was the name of how
what should they Google to find it cont
F context fun right context. fun that's
really excellent analysis of
s1047 um and then I think in in in the
manner of whether or not um this is just
a matter of doing what companies have
committed to do I would have a different
characterization as well um you know I
i' I've been I'm not yet finished I'm
reading the Llama 3.1 paper and candidly
I think meta gave a wonderful gift to
the world with their release of um the
latest llama 3.1
and I think meta actually went through
extensive safety testing um um really
you know and and I feel like but I think
meta feels to me like they're certainly
delivering the spirit of what they're
committed to but the details of that
safety testing I would argue I believe
are quite different than what um some of
the patrol statute um requires and so uh
and and and I agree that 12 already
allows you know people to sue M Builders
hypothetically this is America right
people can sue anyone for
anything I feel like uh I'm concerned
about different Regulators uh creating
new paths to liability just because
there's one way for someone to sue
someone for something does not mean we
need new ways for people maybe attorney
general maybe others to sue people for
other things especially if um uh it
doesn't actually make AI safer uh when
look cons the requirements um I would
have a hard time knowing what exactly I
would need to do to comply with them uh
and then I think um in the man of civil
penalties versus criminal penalties I
think no one wants to commit perjury at
least I hope not but the act of asking a
developer to sign under penalty of
perjury a guarantee of statements that
seem to me as a non-lawyer really vague
before boy before I sign such a
statement I would definitely you know
want to hire lawyers High Consultants uh
the law refers to auditing maybe higher
Auditors and I think all this compliance
mechanism however much it cost um I
don't see how that actually makes AI
safer so I think unfortunately we are at
a turning point where laws that
California or other places can choose to
pass or not pass will have a huge impact
on um the future of Open Source I was
grateful to Chairman Khan's remind
uh before the CER showed up so on the
importance of Open Source and open
weight models and and protecting that
and I feel like this is one of the mo
those moments where um such laws not
just in California but around the world
could have a huge impact on whether
entrepreneurs startups are allowed to
keep on innovating or if we should be
spending our money not on building
software but on hiring lawyers and
hiring Auditors so
[Applause]
I want to mention just one last thing I
was taking notes the S was speaking I'm
I'm grateful that you welcome feedback
and remarks in fact the center I had a
good conversation had a conversation a
few months ago where I shared some of my
concerns about regulating technology
versus applications and as you're
speaking just now has an interesting
thought which is um I'm glad you welcome
suggestions to the to the statute I I
I'm happy to try to do my best to make
suggestions and have to admit
with all due respect that's your job not
mine and I say that with with affection
and respect for the democratic system
I'm I'm just as we spoke a few months
ago I'm happy to continue offering
constructive suggestions I really am so
for example I think anthropic
suggestions to get rid of the frontier
model division that's great I think it
will make the law less bad I don't think
it'll still make it good but it will
definitely make it less bad so I'm glad
you're serly concerning that suggestion
and I candidly struggle to think how to
modify a law that regulates the
technology as opposed to the
applications to make it sufficiently
less vat to support it but but hopefully
we can keep on working on it on on
making it less bad thank
you and that concludes AI startups and
competition thanks for closing us out
Andre thank you Senator weiner thank you
Lena KH thank you Jonathan caner thank
you all the uh Founders who did demos uh
for everybody I'm going to hand it over
to
Gary I just really uh you one of the
things that's pretty clear to me is um
you know we need to do more of these
things in person uh we need to do these
things respectfully uh in front of each
other and to treat each other with uh a
true community and so that's what YC is
thank you so much for coming out here
we're at an incredible moment in history
and time I'm very proud proud to see all
of you thanks for coming out