Liam Conner
The following is a transcript of an interview with Chris Krebs, former director of CISA and cybersecurity expert and analyst for CBS News, which aired on May 19, 2024.
MARGARET BRENNAN: And now we’re joined by CBS News cybersecurity expert and analyst Chris Krebs, who is the former director of the Cybersecurity and Infrastructure Security Agency. It’s good to have you back.
CHRIS KREBS: It’s good to be back.
MARGARET BRENNAN: So in our research that we just did in Arizona in particular, it was a hotbed of voter denial in 2020 and we’re seeing that again. I mean, almost half of Trump supporters say the 2024 presidential results in Arizona should be investigated and challenged if Biden wins. Seven in 10 of those who identify as “MAGA.” Will this be a repeat of 2020? I mean, what’s going to happen?
CHRIS KREBS: Well, you know, all of these allegations occur despite any credible evidence being presented in the courts or, you know, to everyone else and the experts out there. We continue to see complaints of problems with 2020, and until 2022, midterm exams. So I think the incentive structures, the way they’re aligned right now, create a lot of space and a permissive environment for those who claim to be problems, again, despite any credible evidence and continued investments in election security, election resilience , modernization electoral systems. You know, in 2020, we talked about having about 95% of paper votes. Now that number is probably closer to 98. We’ve made improvements in recent years.
MARGARET BRENNAN: It’s interesting that you say that about the role, because on the campaign trail Donald Trump claims that 2020 was rigged. The 2024 claims could be as well. Listen to what he just said.
TRUMP: Our goal in the elections will be to one day vote with paper ballots, proof of citizenship and voter registration. It’s very simple. If you want to save America and your Second Amendment, register, get an absentee or vote by mail, vote early or vote on Election Day, I don’t care.
MARGARET BRENNAN: So, two things there. Now he is in favor of mail-in ballots. He went against them in 2020, but you just said that there is already a documentary trail. He is saying there are none.
CHRIS KREBS: The vast majority of votes cast in the United States right now are documented. There is a small percentage that is registered in the electoral systems, but this is for accessibility purposes, for those who may or may not be visually impaired. And it is important to ensure open and accessible voting processes. The one-day vote, this impulse, this impulse, is actually counterproductive. In fact, it is contrary to participation in the democratic system and to resilience and security. This longer run, weeks, months before the election, allows us to detect if there are any problems, get to the top and solve them. If you had an election in a single day, life happens, first, someone might get sick, they might have a car accident, you might not be able to go to the polls. But if something happens early on, it could have a catastrophic, cascading impact on getting people to the polls and voting on Election Day. So I’m much more in favor of the system we have now, with early voting, with mail-in ballots and absentee ballots.
MARGARET BRENNAN: So the director of national intelligence, your successor at CISA, Jen Easterly, also testified that they are so prepared that the election infrastructure has never been more secure. But we also hear from US intelligence that China, for example, is experimenting with things like generative AI to influence voters. How do people understand this effort to influence versus actually moving the ballots?
CHRIS KREBS: I think that’s the right distinction. There is interference, which is the technical manipulation of systems, and as Jen Easterly, my successor at CISA, said earlier this week at the Senate hearing, the investments we’re making, the continued improvements to the systems, this- I would even suggest that, you know, when, when we said in 2020 that it was a safe and secure election, it’s even safer now, because these continued investments, because the continued improvements in the systems. But when it comes to the influence, the scope, the scale, the technology available to our adversaries, including AI and deep fakes, it’s a much more precarious threat environment. The Chinese are active. Russians are very active. They have been using deepfakes in Europe. We have seen AI emerge in Moldova, Slovakia and Bangladesh. So yeah, it’s going to be a tool. My impression, though, is that AI-powered or AI-enabled threats will be very similar to what happened in New Hampshire with Robocall. It will be detected immediately, it will be investigated quickly and it will be prosecuted. And that’s what’s happening now. I think the biggest concern, though, is that this is cumulative. It’s an aggregator. So, instead of a single AI-enabled catastrophic event –
MARGARET BRENNAN: Right.
CHRIS KREBS: It’s going to be a constant drumbeat where we, where voters, the public, will lose trust and confidence in the overarching information ecosystem.
MARGARET BRENNAN: And we’re already seeing this in our research. A loss of trust.
CHRIS KREBS: Some might say this is part of the strategy.
MARGARET BRENNAN: That’s scary.
CHRIS KREBS: Absolutely.
MARGARET BRENNAN: CNN reported that the Chinese and Iranian governments attempted to use fake AI content in 2020 but chose not to deploy it. Is this accurate?
CHRIS KREBS: I, I can’t speak to the specific intelligence there. It shouldn’t be a surprise. We’ve been talking about deep fakes and AI as a risk vector for years, at least half a decade ago. So to say that, whether it’s the Russians, the Chinese, the Iranians, the North Koreans, whoever, they’ve been tinkering with or researching this technology for just four years, I think is completely credible. We are seeing this now. We are already seeing it, as I mentioned, in Europe. We’re seeing it here too, and I hope that between now and the election we’ll continue to see AI-created content popping up. And this is where it is important for technology platforms to ensure they are aware of this.
MARGARET BRENNAN: Right.
CHRIS KREBS: That AI companies are ensuring that their platforms cannot be manipulated. The challenge, of course, is that there are open source models that can be used.
MARGARET BRENNAN: And we need to continue talking to you about this, unfortunately, over the next few months, because it looks like it’s going to be a persistent problem. We’re back.
