00:00:00
Recently, California Governor Gavin
00:00:02
Newsome signed a bill designed to
00:00:04
strengthen the protections for children
00:00:06
in the online space. The focus of the
00:00:08
bill is around AI and chatbots and how
00:00:11
they interact with kids. To discuss the
00:00:13
bill and its impact, welcome back to the
00:00:15
show Kevin Orbach, who is professor of
00:00:17
legal studies and business ethics here
00:00:19
at the Wharton School. Kevin, great to
00:00:21
see you again.
00:00:22
>> Dan, always great to talk to you. I look
00:00:24
obviously this is a very important topic
00:00:26
when we're thinking about the
00:00:27
development that we're seeing going on
00:00:28
with AI but making sure that we have the
00:00:30
protections necessary out there. So as
00:00:33
this bill is signed and it's now being
00:00:35
put in action. What are your thoughts
00:00:36
about the importance of having something
00:00:38
like this? There are a whole series of
00:00:42
AI related laws that are under
00:00:44
consideration in states throughout the
00:00:45
US and especially in California which is
00:00:48
where this particular bill was adopted.
00:00:52
uh California has really been the the
00:00:53
central battleground because it's such a
00:00:55
big influential state and because
00:00:58
Congress has not moved forward on
00:01:00
federal legislation. So this particular
00:01:03
bill comes about at a time where there
00:01:06
is growing concern about AI companions
00:01:10
and about children in particular being
00:01:12
impacted in some cases uh causing self
00:01:15
harm or even suicide because of their
00:01:18
interactions with AI chatbots and
00:01:21
companions. And so it's a bill that that
00:01:23
actually does less than it sounds like
00:01:26
at first, but at least tries to put a
00:01:30
stake in the ground of saying that the
00:01:32
state of California wants to put
00:01:35
obligations on companies, especially the
00:01:37
ones that are knowingly providing these
00:01:39
services to children. So, you bring up
00:01:41
an interesting point at at the beginning
00:01:43
of that last statement and whether or
00:01:45
not we're talking about this from the
00:01:46
federal perspective or the state level
00:01:48
and it feels like what we need to see is
00:01:51
a lot more push from the state level
00:01:53
because the federal level just doesn't
00:01:55
be isn't able to, you know, move a a
00:01:58
significant piece of legislation forward
00:02:01
to really wrap the entire country in
00:02:03
this. This has been the subject of
00:02:04
tremendous debate. There was a push a
00:02:07
few months ago to have a federal
00:02:08
moratorum where the federal government
00:02:10
would actually prohibit states from
00:02:12
regulating AI for 10 years out of
00:02:14
concern that comes from uh conservatives
00:02:17
and from the AI companies that 50 states
00:02:20
regulating in complex inconsistent ways
00:02:22
is going to be a real burden on
00:02:24
innovation. And there's some truth to
00:02:26
that concern. But there's also some
00:02:28
truth to the concern that you raised
00:02:29
that these are real issues. States are
00:02:32
the laboratories of democracy. They're
00:02:34
closer to the ground. they should be
00:02:36
taking action to protect their citizens.
00:02:38
I ultimately think we need a balance. I
00:02:40
think it doesn't make sense to have 50
00:02:42
different state laws, but it also
00:02:45
doesn't make sense to say we should just
00:02:47
not do anything on these issues. So
00:02:49
really the question is let's look at the
00:02:51
legislation that's been adopted and
00:02:53
assess it on its own merits. And I guess
00:02:55
when you're talking about children and
00:02:57
we obviously we know uh you know how
00:02:59
much the internet is used by kids on
00:03:01
different platforms, different websites,
00:03:04
but you're talking about individuals
00:03:05
that are in their development process,
00:03:07
whether it be their physical
00:03:09
development, but also their mental
00:03:10
development as well. As you kind of
00:03:12
alluded to, this is a very important
00:03:14
subsect of the population that you
00:03:17
really do need to look out for.
00:03:20
The challenge is how do you look out for
00:03:23
them? And this is a debate we've been
00:03:25
having increasingly in recent years
00:03:27
about social media that there is a law
00:03:29
called Kappa that's really mainly about
00:03:31
privacy but has to do with online
00:03:33
services that know they're dealing with
00:03:35
children but it doesn't require strict
00:03:38
age verification. So if you tell
00:03:40
Facebook that you're 13 years old, they
00:03:42
have to do something. But if you don't
00:03:44
or if you're a 13-year-old that pulls
00:03:46
the drop down and says I'm 20 years old,
00:03:49
um then potentially they're not
00:03:51
obligated. We have the same issue with
00:03:53
AI and there's all kinds of complexities
00:03:55
around implementing those age
00:03:56
restrictions but absolutely children are
00:03:59
a distinct category and the AI companies
00:04:02
themselves open AAI anthropic and so
00:04:04
forth have acknowledged that they need
00:04:07
to step up more and that you know
00:04:10
there's one set of issues for adults
00:04:12
thoughtfully using these AI tools.
00:04:15
There's issues even with adults with
00:04:17
with what's called AI psychosis that
00:04:18
some small percentage of people it
00:04:20
actually does drive them crazy. And so
00:04:21
we need to address that. But with kids
00:04:23
definitely uh I think whether it's
00:04:26
mandated by law or just appropriate good
00:04:29
governance and actions by the companies,
00:04:31
they need to take some steps to address
00:04:33
the issues with kids using these
00:04:35
technologies. But there there are some
00:04:37
out there that also say that if you have
00:04:40
a higher regulatory kind of framework
00:04:42
and and world around things like the
00:04:45
internet that it's a way to kind of cut
00:04:48
back on innovation. How do how do you
00:04:50
respond to those those comments? It's
00:04:52
not all or nothing. That's what I said
00:04:53
before. People always want to say this
00:04:56
law will stop innovation and it'll be a
00:04:58
chilling effect. And that's a legitimate
00:05:00
concern, but it's a problem to say
00:05:03
anytime any law gets passed because the
00:05:06
reality is all these companies are
00:05:08
investing literally now trillions of
00:05:11
dollars building this AI infrastructure
00:05:14
because they're competing against each
00:05:15
other. They're competing against China.
00:05:17
They believe they are building something
00:05:19
that is going to be absolutely
00:05:21
foundational to not just the future of
00:05:23
the internet, but the future of business
00:05:24
and the world. And so, yes, there's an
00:05:28
impediment and some cost that they have
00:05:30
to put some guard rails on for kids, but
00:05:32
I really don't think uh companies like
00:05:34
OpenAI and Google and so forth are going
00:05:36
to shut down what they're doing just
00:05:37
because of those costs.
00:05:38
>> So, where do you think we stand right
00:05:40
now? And and maybe even more so, where
00:05:42
do you think we're heading in terms of a
00:05:45
regulatory framework on on obviously the
00:05:48
internet, but also around AI?
00:05:51
Well, that's a big question, but I I
00:05:53
think that honestly the big pressure is
00:05:56
not regulation. The big pressure is
00:05:59
liability. Uh, one of the things that
00:06:01
this law SB243 does is it does create
00:06:05
private liability if companies are
00:06:07
required to take steps to protect kids
00:06:10
and fail to do so. But even without that
00:06:12
law, there are a number of lawsuits that
00:06:14
have already been filed by parents of
00:06:16
kids that committed suicide after using
00:06:18
some of these AI chatbot and companion
00:06:20
tools. And they're doing it under
00:06:22
general principles of tort law. And so I
00:06:24
think uh regulation is important, but
00:06:27
again, at the end of the day, most of
00:06:29
these major companies acknowledge they
00:06:31
need to take appropriate steps. Most of
00:06:33
the enterprises who are the ones that
00:06:35
I'm mainly working with that are all
00:06:36
deploying AI acknowledge they need
00:06:38
appropriate AI governance. And the big
00:06:41
club that they're worried about is a
00:06:43
lawsuit that's going to have massive
00:06:45
amounts of damages.
00:06:46
>> Kevin, always great to talk with you and
00:06:48
get your insight. Thanks very much.
00:06:50
>> My pleasure.
00:06:50
>> Thank you. Kevin Warbach, professor of
00:06:53
legal studies and business ethics here
00:06:55
at the Wart School.
