-9999

Floor Speech

BREAK IN TRANSCRIPT

Ms. KLOBUCHAR. Mr. President, I rise today to again call on my colleagues across the aisle to take action to put in place commonsense guardrails to address the risk that artificial intelligence poses to our elections.

There are some incredible innovations that we are seeing and will see as a result of AI, and our country is in the lead. I want those innovations to be good, and I want us to lead the world as we have for so long, but if we don't put some commonsense rules in place, such as the bill Senator Thune and I have, to create a regulatory framework for nondefense applications or some of the other bipartisan bills that have come out of the Commerce Committee, I think that we are going to lose the positive for the negative. So I am trying to look at some commonsense rules, and one of the most obvious ways that we can put some rules in place is when it comes to democracy, which, to me, is kind of a hair-on-fire moment.

Like any emerging technology, as we know, AI brings both opportunity and uncertainty. As columnist David Brooks has said, the people in AI seem to be experiencing radically different brain states all at once.

He said:

I've found it incredibly hard to write about AI because it is literally unknowable whether this technology is leading us to heaven or hell.

So we want it to lead us to heaven, and we want to get those new cures for medical diseases, and I have seen the work done right in my own State at Mayo.

But one of the things that we must do--and Senator Schumer and Senators Heinrich and Rounds and Young have been leading the way--is to find a group of bills that we could actually pass, and perhaps we could do this at the end of the year. The ones that I would like to see today--which I know cannot happen as we work on the budget but at least by the end of the year--is to do something when it comes to democracy and the kind of videos and fake robocalls and things that we are seeing that could be a dangerous--and have already been a dangerous--threat to our democracy.

Maybe the first time this kind of hit the national conscience was something that wasn't actually created by AI. It was actually the voice. It turned out to be the voice of a magician who pretended that he was President Biden; and in the New Hampshire primary, he called people and said that they should not vote. That case, fortunately, has been taken up in New Hampshire by the prosecutors and is being investigated and prosecuted.

These are the kinds of things, when in the hands of AI, are going to get very dangerous because we are not going to know whether or not it is Donald Trump or Kamala Harris. We are not going to know if it is really them or if it is a fake. We are already seeing videos and voices and things, so I am asking my colleagues to join with me to do what people have already done in so many States, including in red States.

Texas--Texas--has a ban on deepfakes. Minnesota has a ban on deepfakes. States like Mississippi and Utah have put in labeling requirements so that at least you know if something is being created by AI; so when an innocent citizen looks at the video or looks at the ad, they see that at the bottom.

And not everything is the same. If you use AI to, say, change your hair color or do something to make someone look better, that doesn't come under at least the bills that we have for Federal elections that we are proposing. But if you actually are trying to mislead voters about whether or not it is actually the person, that would come under the bill that Senators Hawley and Collins and I--yes, that is right-- that Senators Hawley and Collins and Coons and I and a number of others have introduced that would simply say: Once you do something like that, then you have to take it down.

We have a number of platforms, including open AI and Microsoft and others, that are supporting this bill. This is not some kind of radical idea. We haven't had any serious pushback for this particular bill because we were so careful in drafting it. We understood that we couldn't include, say, satire. When you have a satire, a joke thing that comes out, the Constitution says you can't really ban that, right? That's speech.

But for those kinds of products, we think you should at least say on it--which is allowed; this is what all these States have done-- ``produced by AI.'' That is the bill that Senator Murkowski and I have--the Republican across our aisle, my friend from Alaska. So these are both bipartisan bills, and I am urging my colleagues to look at them.

Election day is now less than 45 days away. We have voters already casting their ballots, and absentee ballots have been mailed out. Truly, this is not a Democratic issue or a Republican issue. This is an American issue. People should be able to know that, when they see a candidate, it is the candidate.

We have a number of bills, as I said, that are before us, and I want people to know--and my colleague from Georgia, the Acting President pro tempore, is well aware--of how people can try to mess with elections. His words would probably be more straightforward than any words I will say today when it comes to trying to stop people from voting or feeding them with information. Sometimes it just comes down to some people don't want some people to vote. That is what that radio ad was about-- radio; I am sorry--that call that was made in New Hampshire was all about: trying to get people not to vote or to think things are so bad that they don't want to vote for the person.

But there is something even more insidious going on, and that is that we have heard from our intelligence Agencies that foreign adversaries, like Iran, China, and Russia, are trying to interfere with our democracy.

Just this month, the Office of the Director of National Intelligence issued a public report confirming this. In May, the Director of National Intelligence, Avril Haines, testified about an increase in foreign influence efforts and how AI can make it easier to amplify deceptive content. By the way, this has happened before in our elections, and it has happened around the world. It is not necessarily, in some cases, directed at the right or the left. The foreign interest just decides: Well, what do I want to mess around? Sometimes, as in the past, Russia just wants to mess around--mess around with the right or with the left--just to create chaos.

FBI Director Christopher Wray echoed this concern when he said that the U.S. will face more adversaries moving at a faster pace enabled by new technology. And the Department of Justice took action 3 weeks ago to disrupt Russian Government-backed efforts to interfere in our election, which includes the use of AI to spread disinformation online.

We have seen this technology being used to generate viral misleading content to spread disinformation and deceive voters about candidates from both parties.

In July, as I note, there was just recently the deepfake of Vice President Harris's voice saying things she never would say about President Biden. According to reports, it was seen more than 130 million times in just 4 days. Now, you know, maybe if you were to watch the whole thing, you would know that it is not her. Not everyone does that, right? You look at snippets on the internet; you look at part of it. Again, I had people say ``wow'' to me. I know that seems strange for people who are in the Beltway, and they watch everything, but if you just watch something for a few seconds, you may not know whether or not it is actually the candidate or not. That is why, in this case-- because it was arguably satire--you would at least have a label through the whole thing, and that would be a requirement if we pass the bill Senator Murkowski and I have.

Earlier this month, Taylor Swift talked about how AI was being used to make it look like she endorsed a candidate that she didn't even endorse. As the election approaches, now is when disinformation can have its biggest impact and when we could possibly see the worst of the worst.

So whether you are Democrat or Republican, no one wants to see fake ads or robocalls when you cannot even tell if it is the candidate you love or the candidate you don't like. That is why, as I note, 19 States across the country--red, blue, and purple--have passed laws in this area.

On the Federal level, the bill I have with Senator Collins and Senator Hawley is sitting there. We got it through the Rules Committee. But unfortunately our colleagues on the other side of the aisle have blocked this bill from being considered in the past when I asked unanimous consent for it.

I note it is a bipartisan bill. It is supported by 40 national security experts and current and former government officials, including former Secretaries of Defense Chuck Hagel and Leon Panetta and Secretaries of State from both parties. It is supported by former Federal Election Commission Chairman Trevor Potter, who was appointed by a Republican President. I mentioned OpenAI, Microsoft, IBM, and Salesforce supporting the bill. It has a bipartisan companion bill led by Representatives Derek Kilmer of Washington and Tony Gonzales, a Republican from Texas.

We must get this done.

At this point, we are going to have to now rely on the platform--some of whom have policies in place--to take down these fake ads and these fake videos. Many of them have committed to do so. But those are not all the platforms, and we are much better off if we have standard legislation across the country that doesn't preempt the State laws for their own work but actually sets a standard for Federal elections-- President, Senate, and Congress.

So I hope this isn't as bad as I think it is going to be through October, but I think we are already seeing signs that it is. That is not democracy. We cannot have a functioning democracy if our people can't tell if it is their candidate or not. It is going to make things much, much worse.

We can't stand on the sidelines. Let's work together for the sake of our democracy to put in place commonsense rules of the road on AI to uphold trust and faith in our election. I stand ready to work with our colleagues in any way to get this done.
BREAK IN TRANSCRIPT

Source