[Congressional Record Volume 170, Number 125 (Wednesday, July 31, 2024)]
[Senate]
[Pages S5654-S5655]
From the Congressional Record Online through the Government Publishing Office [www.gpo.gov]
Unanimous Consent Request--S. 2770
Ms. KLOBUCHAR. Madam President, I thank the leader for his support
for these bills, but also for his bipartisan work on artificial
intelligence, including the group that he put together with Senator
Heinrich and Senator Rounds and Senator Young to really lead a
bipartisan effort. One of the major focuses of that effort has been
doing something on the democracy front.
While some of these bills actually passed through the Commerce
Committee today on a bipartisan basis to start the initial work of some
sensible guardrails on AI, the democracy work actually can't wait
because, as the leader pointed out, we are less than 100 days from this
election; and we are seeing States act across the country--red States,
blue States, purple States--putting some simple rules in place for this
new sophisticated technology.
I always believed that our laws have to be as sophisticated as those
that are trying to mess with them, and we cannot simply stand by on the
Federal basis for the Federal elections--we are not talking about
messing around with their State rules. They are doing their own rules.
We are talking about Federal elections when it comes to involvement
with AI.
AI, as we know, is set to become one of the most significant
technological advances of our time. Like with any emerging technology,
it brings tremendous opportunities, but it also brings tremendous risks
and uncertainties. I think David Brooks, the columnist, put it well
when he wrote:
The people in A.I. seem to be experiencing radically
different brain states all at once. I've found it incredibly
hard to write about A.I. because it is literally unknowable
whether this technology is leading us to heaven or hell.
Well, it is on us right now as the elected representatives of the
people of this country to make the decision of what fork are we going
to go on. If we put no guardrails in place when it comes to scams, when
it comes to messing around with people's intellectual property rights,
when it comes to national security, when it comes to democracy--which
is our topic today--then we are not going to unleash the potential and
the great opportunities of AI because we will not have put the
guardrails in place to make it safe.
This means protecting ourselves from the significant risks AI poses
without stifling innovation and working to preserve trust in business,
government, and our elections, as we all adapt to this rapidly
advancing technology.
With this year's election so soon in front of us, we must put in
place these commonsense rules. We have heard repeatedly about the
potential of AI to upend our election. All of our witnesses from both
parties agreed that this was a threat when we had our hearing. And at
the bipartisan AI forum that I just mentioned, there was consensus that
Federal legislation is necessary; that disclaimers are not enough in
some cases; and that it is critical to our national security.
By the way, these AI videos or fake robocalls or videos of people
that aren't really the candidates that you don't like or you do like--
if you don't know who you are watching, how are you going to be able to
make your decision as a citizen in this great democracy?
And, by the way, these could be promoted by foreign governments, by
foreign countries. We have seen this in Canada where they just
completed an investigation and found that China had meddled in their
elections, in their elections for parliament seats.
This is happening right now, and we need the ability to take these
things off or at least label them so people know what they are viewing
and what they are listening to. This is a hair-on-fire moment.
AI has the potential to turbocharge the spread of disinformation and
deceive voters. This is happening to candidates on both sides. In the
New
[[Page S5655]]
Hampshire primary on the Republican side, a video was released with
fake AI-generated images of former President Trump hugging Dr. Fauci.
That wasn't true. We have seen AI being used to generate viral
misleading content about our colleagues in this Chamber, including a
fake video with Elizabeth Warren--that wasn't really Elizabeth Warren--
telling people that she didn't think that Republicans should be allowed
to vote. Complete lie. It wasn't her. But it looked like her and talked
like her.
We have seen this all over the country, and that is why States have
been acting; 18 States across the country have already passed laws in
this area, including my home State of Minnesota, which banned deepfakes
of candidates 90 days before an election. Texas has a ban on deepfake
videos of candidates. And that passed unanimously; the Minnesota bill,
I think, one person voted against it. Democrats and Republicans joined
together to say: We are not going to have these deepfakes because they
could happen on either side, and our citizens aren't going to know who
they are looking at and if it is the real Donald Trump or if it is the
real Kamala Harris or if it is the real Amy Klobuchar or the real
Senator Fischer.
Other States who have done something on this: Alabama--these have
been mostly disclaimers--Alabama, Arizona, California, Colorado,
Florida, Hawaii, Idaho, Indiana, Michigan, Mississippi, New Mexico, New
York, Oregon, Utah, Washington, and Wisconsin. If you listen to those
States, you are not like, Oh, those are all blue states. Oh, those are
all red states. Those are Governors and legislatures that decided we
cannot just take this as it is not going to be a problem, and it is all
fun and games. They have decided that: We have got to make sure our
citizens know, for State political advertising, what is going on here.
Some tech companies are also taking action because they know that
this technology has a potential to sow chaos in elections, but we
cannot rely on a patchwork of State laws for just about half the
States--probably what it will end up being--and voluntary commitments,
as important as those are.
That is why as chair of the Rules Committee, we held a markup in May
where we passed three bipartisan bills to take this head on.
I am calling on the Senate today to pass, first of all, the
bipartisan bill with Senator Hawley, the lead Republican with me on
this bill; with Senator Coons, Senator Susan Collins, Senator Bennet,
and many others to ban AI-generated deepfakes of Federal candidates,
within the framework of the Constitution.
So what does that mean? Well, that means an exception for parody and
satire as well as reporting by news organizations. So we drafted this
bill with Democratic and Republican lawyers in a way that it could be
upheld in court under the Constitution.
Our bill is supported by a bipartisan group of more than 40 national
security experts and current former government officials, including
former Secretaries of Defense Chuck Hagel, a Republican, and Leon
Panetta, a Democrat, and Secretaries of State from both parties. It was
also endorsed by the former Republican Chairman of the Federal Election
Commission, Trevor Potter, as well as tech companies like OpenAI,
Microsoft, IBM, and Salesforce.
These companies actually want to be able to say: This is a deepfake.
It is not the actual candidate, and there is a law that says we have to
take it down.
That is what this is about, as well as allowing the defamed
candidate--the person who it is not really the person in the video or
the ad or the robocall--to sue whoever has done this to them. That is
the way in other areas in our law we are able to stop bad conduct.
In the House, a bipartisan companion bill is led by Representative
Derek Kilmer of Washington and Tony Gonzales of Texas.
With election day approaching, we have the opportunity to come
together on a bipartisan basis to counter the threats that AI poses to
our elections and protect public trust and faith in our democracy.
Now, there is a second bill that I will call for in a moment that
deals with things that are maybe parody or things that don't rise to
the level of the deepfakes or are in a different category that could
complement this bill as well. But this is for the worst of the worst.
And that is why we have had strong support from a conservative like
Senator Hawley, who certainly is aware of what the Constitution says
and what our rights are; moderate Republican like Senator Susan
Collins; and many others to support this bill. So now I will call for
this bill.
As if in legislative session and notwithstanding rule XXII, I ask
unanimous consent that the Senate proceed to the immediate
consideration of Calendar No. 388, S. 2770; further, that the
committee-reported substitute amendment be agreed to; the bill, as
amended, be considered read a third time and passed; and the motion to
reconsider be considered made and laid upon the table with no
intervening action or debate.
The PRESIDING OFFICER. Is there objection?
The Senator from Nebraska.
Mrs. FISCHER. Madam President, reserving the right to object, could I
have the Senator please clarify which bill she called up.
Ms. KLOBUCHAR. Yes. This was the bill--
Mrs. FISCHER. 2770?
Ms. KLOBUCHAR. 2770, which is the deepfake bill. The other bill is
3875, the Klobuchar-Murkowski bill, which is the disclaimer bill.
Mrs. FISCHER. Thank you.
Madam President, like many of my colleagues, I am concerned about
artificial intelligence-generated deepfakes in the context of political
speech and election administration. But the Protect Elections From
Deceptive AI Act is not a solution to this problem. The bill recycles
provisions from the partisan For the People Act. It is overly broad,
and it would prohibit the distribution of political ads that include
AI-generated audio or visuals, including commonly used image and video
editing programs. It greatly expands the regulation of protected speech
and uses vague terms that will inevitably chill that speech.
This bill does not balance First Amendment rights with the evolving
challenges that we have with the digital age; and, therefore, Madam
President, I object.
The PRESIDING OFFICER (Ms. Cortez Masto). The objection is heard.
Ms. KLOBUCHAR. Madam President, I will note again that Senator Hawley
is the lead Republican on this bill, and I don't think for a minute
that he would allow for a bill that is something that he disagrees with
when it comes to being radical or broad or anything like that. He
simply agrees with me, as does Senator Collins, that we have a major,
major issue here with these deepfakes that are going to extend to, say,
what happened in New Hampshire, which involved a fake robocall ad from
President Biden that people believed was him calling on people not to
vote. That case is, of course, being investigated by a Republican
Attorney General, and a case is being brought in New Hampshire.
But the point is this is just the beginning. We haven't even entered
the general election yet. That will start at the end of August. And so
that is why time is of the essence here, and that is also why we
drafted this bill with Democratic and Republican lawyers in a method
that was narrowly tailored so that it would abide by the Constitution.