WASHINGTON AI NETWORK PODCAST TRANSCRIPT Host: Tammy Haddad Ep. 15 University of Virginia Center for Politics 25th Anniversary

Guests: Jen Easterly, Director of the Cybersecurity and Infrastructure Security Agency (CISA), Georgia Secretary of State Brad Raffensperger, and UVA’s Kyle Kondik, Managing Director of Larry Sabato's Crystal Ball.

Tammy Haddad: [00:00:00] Welcome to the Washington AI Network podcast. I'm Tammy Haddad, the founder of the Washington AI Network, where we bring together official Washington, DC insiders, and AI experts who are challenging, debating, and just trying to figure out the rules of the road for artificial intelligence. This AI revolution has been led by industry and governments are trying to catch up. The conversation you're about to hear took place at the 25th anniversary celebration of the University of Virginia's Center for Politics. Many of you follow Larry Sabato's crystal ball. In this episode, you'll hear from two of the leading forces for fairness, safety, and security in the 2024 election. My guests are Jen Easterly, Director of the Cybersecurity and Infrastructure Security Agency, Georgia Secretary of State Brad Raffensperger, and Kyle Kondik, Managing Director of Larry Sabato's Crystal Ball. Join us as we [00:01:00] discuss how artificial intelligence may impact the 2024 election. But before we get started, I hope to see you at the Special Competitive Studies Project. AI Expo for National Competitiveness. It's May 7 and 8 at the Convention Center in DC. No matter what you're doing in AI, technology, or government, you don't want to miss it. We'll also be there with podcast episodes, new ones, the Washington AI Network podcast from the heart of the Expo. See you there and thanks for listening to this important conversation. Announcer: [00:02:00] And now from the University of Virginia’s Center for Politics. Emily Horn: Good afternoon and welcome to this kickoff event for the Center for Politics 25th anniversary weekend. My name is Emily Horn and I am a student intern with the Center for Politics. Now it is my pleasure to introduce the moderator for today's panel Tammy Haddad, president and CEO, Tammy is a media innovator whose Washington DC company devises winning strategies for clients, including some of the world's top media companies, technology, disruptors, innovative startups, and nonprofit organizations. She is the veteran television executive and executive producer with more than 25 years of experience for many landmark web, cable, and network programs, including [00:03:00] The Today Show, The Late Show, and Larry King Live. She is the co host of Bloomberg's Masters in Politics podcast, interviewing the biggest names in politics, media, and policy. A longtime Washington partner for HBO, she is a consultant on the HBO hit comedy series, Veep. And on HBO Films, including confirmation and all the way. Haddad is the co founder and editor in chief of WHC Insider, the White House Correspondents Weekend Insider website covering the political power and media cultures of Washington, D. C. And she plays a principal role for Washington's most important events, including the annual White House Correspondents Weekend. Would you please join me in welcoming our moderator, Tammy Haddad. Tammy Haddad: Thank you all for coming today to this very important panel. I couldn't be more thrilled. These are the right guests. They know everything. What they'll [00:04:00] tell us, we don't know. We're not sure. You guys know Brad Raffensperger,. I was in the back with Brad. I'm like, when you're done with this stuff, come and be a TV person. Come on now. You know how much fun it is. Thank you, Brad, so much. And Jen Easterly is a remarkable person. She's in charge of security, safety, cyber, everything for elections. So critical. It's such an honor to be here with you. I watch you work. I have her Google alerted. If you don't, you should, because if you want to know what's going on in this election, all these elections, she's the person, Jen. Tammy Haddad: Thank you so much for that. Jen, we do have to start with you. We are talking about AI in elections. First of all, everyone, how many here have used AI? Chat, GPT? Raise your hand. Jen, I know you're dealing with this issue for the elections, but let's just ask you straight up your biggest concerns about AI. Biggest threat. Tammy Haddad: Jen Easterly, Director [00:05:00] of CISA. Jen Easterly: Can I just frame it a little bit? So first of all, CISA, I don't know if anybody, everybody's familiar with it, but just to give the elevator pitch. So CISA is Cybersecurity and Infrastructure Security Agency. If you know about it, you probably know about it because of my predecessor, Chris Krebs who was the first director, obviously, UVA alum, I think, on the board for the Center of Politics. Tammy Haddad: Larry's student. Tammy Haddad: Larry Sabato, the University of Virginia Center for Politics. Jen Easterly: Larry's student, indeed. CISA is the newest agency in the federal government. It was stood up at the end of 2018 to play two key roles, to be America's cyber defense agency, As well as the National Coordinator for Critical Infrastructure Security and Resilience. So the mission is to lead the national effort to understand, manage, and reduce risk to the cyber and physical infrastructure that Americans rely on every day to get water and power and transportation and communication and health care and education. [00:06:00] And in the aftermath of the 2016 election, where the Russians attempted to manipulate the presidential election infrastructure, which before had not been considered critical infrastructure, was actually designated as critical infrastructure at the beginning of 2017. And CISA was named as the federal government lead. In the role of working with state and local election officials, who, of course, are on the front lines of defending our election infrastructure from the full range of threats, cyber, physical, operational, foreign activities. So that's the role we play. We're here to help state and local election officials with resources, capabilities, threat information. Now, just three key points, and I promise I'll get to your question, but just to frame the discussion. First off, since election infrastructure was designated as critical, there's been a whole ecosystem that has grown up around the election stakeholder [00:07:00] community. Obviously, state and local election officials who are on the front lines of defending democracy, but the federal government is now involved in providing information, not just CISA, but FBI and DOJ and the intelligence community to understand and share information about threats. You also have election vendors who are focused on the security of election equipment. You have the private sector that can provide technology resources down to election officials, and then the local law enforcement non profits like the Election Infrastructure Information Sharing and Analysis Center. So they share real time threat information with all 50 states, 38 3, 800 jurisdictions. The thing that I think is really important to understand about this very solid ecosystem that has evolved is that since the federal government has been tracking elections 2018, 2020, 2022 election officials have run safe and secure elections [00:08:00] to be, yes, exactly right. And Secretary Raffensperger can talk at length about this, but to be clear, there is no evidence. That any malicious actors altered, deleted, or changed any votes that had any material outcome on the election. Okay, this has been validated time and time again, including in multiple court challenges and in places where it was close in the 2020 presidential election, there were paper ballots, which were counted and recounted. So I think it's just important to understand, and you'll hear from one of our election officials that they work tirelessly, and they don't look at election infrastructure security as a partisan political issue. This is all about security. And I've had the privilege to work with election officials like Secretary Raffensperger, and I know how tirelessly they work to ensure that all of their citizens votes are counted as cast. And that's why I have confidence, [00:09:00] notwithstanding the threats, that's why I have confidence in the security integrity of elections. And that's why the American people should have confidence in the security and integrity of elections because of election officials like the Secretary and those across 8, 800 jurisdictions. Point one. I'll get quickly to the two and three. Tammy Haddad: That's a critical point because point one is it's people that run elections. It's not AI taking over. Jen Easterly: Exactly. And it really is down to state and local election officials. So point two, the threat is real. And as opposed to 2020, when we were very focused on cyber threats, now you've got cyber threats, you've got physical threats of harassment against election officials, which is really a horrible thing, as well as polling places, you have insider threats, and you have the very real threat of foreign activity. manipulation, interference, influence operations from China, from Russia, from Iran, and all of this accelerated by generative [00:10:00] AI. So generative AI being capabilities that can create original text from pre trained statistical models that use patterns using pre existing tech. So it can create deep fakes, can create all sorts of fake calls, fake video. So that's a very real thing, but at the end of the day, I'll give you the top line assessment. AI will not fundamentally transform the threats to election infrastructure. Those threats exist. It will exacerbate and intensify them. And point three, it is up to all of us to ensure that we secure elections, whether it's state and local election officials, federal government, the private sector, election vendors, generative AI companies that need to show that they can validate when people are using AI in a deceptive way in whether it's campaigning or whether it's for election election voting processes. And then finally, what I always [00:11:00] say, and I'm sure you'll hear this reinforced, if anybody does not have confidence in election infrastructure, not only should you vote, but be a poll worker. Be an election observer. Talk to your state and local election officials, because at the end of the day, they are the authoritative sources, they are the trusted subject matter experts, and they are the ones who can sit down and explain to you how elections work. There is a real threat, but it takes all of us to secure the elections, and I have confidence because of people like the Secretary. Tammy Haddad: Brad, we've got to come right to you. Tammy Haddad: Georgia Secretary of State Brad Raffensperger Brad Raffensperger: And? Tammy Haddad: How are you handling it? I know that you've got this legislation. And you've been working since the last election. Take us through where you are on AI in elections and what you're recommending. Brad Raffensperger: First off, you have to understand that AI does keep you up at nights. In fact, Wall Street Journal had an article today about China. And their threat really is what they're trying to do is separate people. And you have right now a very polarized electorate. [00:12:00] And what they're just trying to do is magnify the polarization. To get us arguing with each other, so we understand that, but in Georgia, what we now have is photo I. D. For all forms of voting. First of all, we have a paper ballot for all forms of voting, so that means we can do 100 percent ring recount if we have to, like we did back in 2020. We have photo ID for all forms of voting. Probably 6 percent of Georgians will actually vote absentee. About 60 to 65 percent are going to vote on one of the 17 days of early voting. And then also, then election day, about 30%. All with photo ID. We also have done health checks, which is what we called that. We went out to all 159 counties, did a sample set of the equipment, made sure nothing had happened to it. Then we also brought in our federal partners. DHS looked at the physical structure of the building. In other words, is this actually a safe and secure facility that you have where this equipment is stored? So then we did testing on the equipment, and then we had reports, and we did that for all 159 counties. [00:13:00] Then, before we have an election, we do logic and accuracy testing, again, to make sure that the equipment is accurately recording your vote, so nothing's inaccurate. Tammy Haddad: And who does that? People that work for you? Brad Raffensperger: People that work for us. And then, for the presidential primary, for the very first time, we believe in America, we actually did what's called parallel monitoring. So we went out to a random county and we just grabbed a random piece of equipment, brought it back to where we do this. Election day processing, which is a secure facility. We don't tell you where it's at, but it's a state facility, but we then did parallel monitoring. We just verified that equipment hadn't been somehow tampered with to give people additional confidence. Then after the presidential primary, it was pretty obvious who was going to win the Democrat race and the Republican race. We did an audit of that, and it turned out that I believe President Trump got five extra votes, Nikki Haley lost one vote, and Ron DeSantis got plus one. And that's all over, [00:14:00] that's the total for the whole state with 159 counties. So many of the counties, many of the counties just were right on. But what you really understand as Jen was just saying, it's really people having faith in the system. And what people are trying to do is destroy faith in the system. And really what is the glue that holds things together is confidence. And that's the real gold standard. So we want people to have confidence in that. So we're looking at what we can do to shore up confidence. So we are also getting ready now for our May general primary. That's when the state reps, the state senators county commissioners or sheriffs will be on the ballot. And so we're actually building the ballots right now. And many voters don't understand that we have about five counties. We didn't get that last information. So we can build the ballot. For UOCAVA, which is your overseas military, but also anyone that's an American citizen overseas. And so actually with our overseas military ballots, we use what's called rank choice voting so we don't have, it's an instant runoff because we're a runoff state [00:15:00] and that allows them to make their choices so we can report those results so that if we had a runoff, we don't have to worry about how do we get things over there and back with that four week window that we have. So it allows us to have runoffs and we will have runoffs because we have more than two candidates in many of these races. We're just prepared we expect it'll be a very interesting cycle. That's probably an understatement. We understand that it's going to be very competitive, just like probably your state is, maybe not so much, but we just know that people can trust the results. And we've had recent polling, 79 percent of all Georgians trust the results that they have in Georgia. Tammy Haddad: That's Georgia. I have to ask Kyle, what about the rest of the country? Because people are really worried about AI. Tammy Haddad: Kyle Kondik, Managing Director of Larry Sabato's Crystal Ball Kyle Kondik: And look, we're also in a time where people are just so pessimistic about all of our big institutions in the United States. If you look like Gallup does, has done long term polling about people's confidence. And really [00:16:00] the only kind of big institutions or categories in which people have a lot of faith, still in the military and small business as a general kind of thing. But if you look at. Congress, the courts, big business, big tech companies, all sorts of things. It's like 25 percent level of confidence or even far below. Congress, I think it's Tammy Haddad: You know, what used to be number one, FBI? Kyle Kondik: I would imagine that it's Tammy Haddad: 25 years ago. Kyle Kondik: I would imagine that's not the case now. I don't know if they ask specifically about it one of the things that I guess is just I always think about with the sort of all the basically made up stuff that, that ends up out there because of AI or just because of other reasons is that we're at a time where people like really distrust the other side, like the Center for Politics has done polling on that, others have done polling on that, and and so I think there's probably like a lot of inclination to believe the worst and you're going to get fed the worst. Kyle Kondik: Sometimes it's real, sometimes it's bogus, and that's just, I just think it's one of the big problems in the country that we just have to grapple with is just this lack [00:17:00] of faith in so many things and the people that try to exploit it. Secretary Raffensperger mentioned the Chinese, and we think about the Russians hacking in and knowing that's the state of our country and really trying to do their part to make it worse. Tammy Haddad: Jen, you testified and talked about the threat from China. Can you spend a little more time on that now. Jen Easterly: Yeah, sure. In terms of the things that I'm focused on our top priority is election security. Just behind that, and there's a clear nexus here, is keeping the country safe from nations like China, which have showed a real propensity to burrowing deep into our critical infrastructure, not just to steal intellectual property and data, but now what we've seen over the past year is an intent to lie in wait in our critical infrastructure to prepare for disruption and destruction. And that is [00:18:00] in the event of a major conflict in the Taiwan Strait. So this is a world where a war, halfway across the planet, could have real impacts on the lives of Americans here at home, through the disruption of pipelines, the destruction of communication systems, the pollution of water systems, the takedown of power systems and so we've been very focused on working with critical infrastructure to identify places where Chinese cyber actors have burrowed into our critical infrastructure and essentially hunting for them and eradicating them. So a lot of concern about where we could be in the next couple of years if Xi does decide to reunify with Taiwan in a military way. And then in the short term, we're very concerned about Chinese impact on our elections. We have seen, it's been publicly out there in the Intelligence Community Annual Threat Assessment, [00:19:00] both in February, as well as the publicly released version that looked at the 2022 midterms that China is getting more involved in interference and influence in our elections. And so very concerned again about influence operations, against disruption, destruction. So we are working to do everything we can with critical infrastructure to ensure that they are prepared because it's really not a scenario where you can prevent bad things from happening. Like in elections, there's there's always something bad that's going to happen. Somebody forgets the key, there's a storm but you have to be prepared for that disruption. You have to build resilience and so that you can deal with any sort of disruption and then be able to respond and recover effectively. So that's what we're doing working with critical infrastructure. Tammy Haddad: And Brad, didn't you have to deal with China in the, was it in the 20, 2020 elections or 22, specifically targeting Georgia? Brad Raffensperger: I believe so, but also I [00:20:00] think someone said that I had a brother named Ron from, worked with the Chinese and no relation, same last name. Could you believe it? I'm sure we'll have threat vectors from China, and we actually saw Fulton County, which is our largest county in the state of Georgia, they got hit with ransomware. Our IT director knew a few people and started asking some questions and he, when he started not getting a straight answer, we immediately make sure we cut our voter rolls separated so they couldn't have access to our voter registration database. And it took probably about four to six weeks before they've got all the issues solved. We said, okay, you can reconnect now. Meanwhile, we're trying to run the presidential primary. But we want to make sure that our system is secure, so our IT team is really top notch, and they're on top of that. We're also talking about... Tammy Haddad: Do you work with other states? Brad Raffensperger: We work with other states because we're we have the National Associations of Secretary of States. We look at these national issues, global issues together, because [00:21:00] we all have the same factors on that. It doesn't really favor any one party. All of us are concerned about it. So there's, we're we have a meeting of the mind, so we work together on those issues. I just think it's really important that we also internally, we have two factor authentication, change your pass codes. And so we have every week you're going to get pestered to do a little check to make sure you're not getting faked by these things. So it's a way to keep all of our employees on their toes. And it's not just our agency, it's all the other agencies that state government or federal government would have, because they're just looking for a way to worm their way into it. And then they have access. Tammy Haddad: Yesterday, NBC News reported that a Chinese national was arrested for hacking the AI of Google. The FBI director said that Silicon Valley is getting their pockets picked. How are you guys going to deal with that? Jen? Jen Easterly: I just wanted to make a comment, because I love when people talk about two factor authentication. And Tammy Haddad: Does everyone use that? [00:22:00] Jen Easterly: Everybody use MFA? Tammy Haddad: if you use it. Raise your hand if you don't do it. Brad Raffensperger: What is it? Jen Easterly: I can retire now. If you come up to me I'll give you the tutorial on it. But so AI artificial intelligence, generative AI will make phishing more tailored because you can aggregate a lot of data and then you can actually tailor a phishing email, so a legitimate looking email that somebody will click on and then you could get hacked and your data gets stolen. So generative AI is something that will make spear phishing emails more tailored to an individual, but as the secretary was saying multi factor authentication, the basic steps that you take to protect your systems from cyber security are those same things that will prevent generative AI from being successful. So I think That's really important to understand, is there's not some exotic new steps that are taken. [00:23:00] It's all of the technological, the physical, the procedural controls that state and local election officials have already put in place, and that's why I am not as freaked out as everybody seems to be about generative AI. Tammy Haddad: Well, I think they're freaked out because we've read stories and seen poll workers who have been attacked who are nervous that there's so many openings. How are you recruiting people, Brad? Brad Raffensperger: We're actually in fairly good shape in Georgia compared to some of the other bellwether states. I think my perspective is, because I ran for reelection, so did Governor Kemp, is that we really went out and talked to everyone in this, across the state. So we actually could then talk to community groups like this one. Here's what happened in 2020. And we say a few things, maybe 7-10 minutes, take questions. Because we won't answer everything in seven to 10 minutes. And that way we answer everyone's questions. People still may not be pleased with the results. I get that. But they realize, "Oh, okay, you did check this. You did check that." And so I think gave people [00:24:00] confidence that they could move on from that grieving process. You have to understand our state was reliably Republican for over 20 years. And all of a sudden, what happened here? And so people couldn't understand it. So this was a way to work through that process. Here's what happened there. There weren't thousands of dead voters. There were just four. There were lots of underage voters, but actually in Georgia, you can register to vote when you're 17 and a half. We checked every one of those day month year. They were all 18 by election day things like that. Tammy Haddad: You have a lot of people that vote By mail, right? What's the numbers? Brad Raffensperger: We really have so in 2020 we had about 40% for the primary in June, and then we had in the fall about 25%. But when you rolled into 2022, we went back to about 6%, which are hits a historical average. Most people in Georgia like to vote early because it's very convenient. Monday, you have five days for three weeks and then two Saturdays, you got lots of opportunities. You see a line out there, you just drive around and come back the next day or come back two hours later. And [00:25:00] so people can work their work schedules or their day schedules with that. And then 30% show up on election day. But also what we've done, put into state law, lines have to be shorter than one hour. And that is huge because that goes back to confidence. Short lines, "man, they must have their act together." that's a good one. And the other one is we're going to have for this cycle this year, You have to post your results by 8 o'clock, 8 p. m. Florida's at 7. 30. They've been working at this a little bit longer. So that's the gold standard and we're working towards that. But 8 is still great because our bigger counties would sometimes not report and then just dump a whole bunch of ballots. No, now we want you, as you do precinct by precinct, report those so people can just see the number of totals go up. It all is about accuracy, but also confidence. It's not in code about confidence, but it is, I think, an essential element, particularly when we are so polarized. Tammy Haddad: And it's investing all of that money. You're actually investing because it's your highest priority. [00:26:00] Kyle, let's add media into this, right? Because just to Jen's point, people keep talking about the negative side. And by the way, I'm going to ask you about the positive things about AI for this election. But for right now, if your media, how many media here want to go into media any of you? None? Are you guys kidding? They kidding? I think they're kidding. Okay. How would, how as media do you cover all of this and ensure that you're giving people accurate information? You've got the feds, you've got the state officials, they're telling you how you're doing it. And then you're hearing from, maybe it's from China, maybe it's from. Washington, maybe it's from No Labels, somebody that, "Oh it's not exactly what you think." What do you do? Kyle Kondik: Let me take a little detour here. Tammy Haddad: Kyle Kondik, Managing Director of Larry Sabato's Crystal Ball Kyle Kondik: So there was just a big election elsewhere in the world. recently. [00:27:00] In Russia, it's real nail biter. I think Putin got 87%. Russia has a long history of basically bogus rigged elections. And so this is a little over my head, but I read about it a little bit, and statistically, it's pretty easy to prove that election was rigged. You can look at basically that certain places had bogus made up results. One of the things about U. S. elections is that despite Donald Trump or anyone else might say, we don't have rigged elections in the United States. But what if we were to have a rigged election? What would it look like? And I think it would be very difficult to actually do it because a lot of people would immediately say, "wow, these results seem strange in certain ways." And this actually, this not rigging the election, but Results that look strange. So like I've been fortunate enough. It's been really cool to do this last couple of years work. I've been working with CBS in their decision desk, and I've got access to all this great information that Edison Research puts together for the big news networks. And one of the things that happens is that there are results that come through that are wrong. Initially, [00:28:00] there's people type in the wrong numbers. If you remember, During the 2020 campaign, it was Antrim County, Michigan that the Trump people got really upset about because they basically put, they, they put the results in wrong and Biden was winning that county and it's a county we know that Biden would not win. They fixed it, but then it led to this months and months and people are probably still complaining about it now that it was a sign of fraud or something, which I didn't even really get the argument. The thing is that mistakes do happen, but it's also pretty easy to see where those mistakes are and then to inquire about it and see what happened, did somebody put the wrong number in, and the media does play a big role in looking at when results are different and if there is some sort of problem and trying to point it out. And again, I think there's this great article, academic article, written about the 2020 election the miracle and tragedy of the 2020 election and the miracle was that during a pandemic, we ran a great election, I think across the country. But unfortunately, there was so much distrust and misinformation about it, but it was quite a logistical undertaking. A lot of states going [00:29:00] to mail in ballots for the first time in bulk, and then that sort of has fallen away in certain places, but there usually are innocent explanations for things that seem funny. And if there actually was if there were results that seemed way out of character for the way the country votes, I think people would be able to recognize it and be able to try to investigate it. Tammy Haddad: I didn't know you were inside on CBS on the election coverage. Maybe if you have another minute to explain how decision desks work because I don't think people really know. Kyle Kondik: Yeah. So Tammy Haddad: This is inside stuff, guys, right? The students should be writing this down. Kyle Kondik: Yeah, when I first I said something about it on Twitter and Natalie Jackson is a polling expert. Is a scholar to center for politics. It's like the first rule of the decision desk is that you don't talk about the decision desk. Or whatever. But, so I mean we're getting a lot of the information is Conan the New York Times website or whatever there's a backend system and there are lots of statistical models. Frankly, some of the sort of models that were used in the past to try to project these elections. They don't [00:30:00] quite work as well. Part of it is because of this intense polarization between vote method, which is new. It's a 2020 thing, particularly a state like Pennsylvania, for instance, like we know the mail in ballots, which are relatively new thing in Pennsylvania, they're going to be super democratic. And then the election day vote is going to be more Republican leaning than whatever the final result is. And so there's the sense I've gotten, even though I'm been new to it just past couple of years, is that it's a lot harder than it used to be. To be able to like really feel confident in the calls. And I'm fortunate at CBS there's a lot, there's a real high bar for accurate for making sure that they're accurate or whatever. But yeah, it's a lot of fun to do, but it's also, it's a great responsibility. And frankly I'm not the one who actually has to make the calls myself or whatever, but the higher ups the full time people there do that. But that's a, that's an important piece of it. Tammy Haddad: When I was political director of MSNBC, and I had more awareness because I'm more of a show person, right? You get the information in the show. And it was a surprise that these models were chosen in [00:31:00] advance. And it's one of the biggest issues this year. You could be wrong. It's a reason why you're wrong sometimes, and the demographics of course change. Brad, you've been dealing with all of this for a long time, expectation game, telling people's story, you're a great communicator. Okay, I'm going to turn to deepfakes, and I'm coming to you first. Are you worried about deepfakes? How are you guys handling that in Georgia and what's your plan? Brad Raffensperger: Our plan Tammy Haddad: I don't mean deep fakes about you. I mean for the presidential race. Brad Raffensperger: We really lean a lot, because they have resources that we just don't have, to DHS and CISA. They just have additional capacities that we don't have. And then we look at what GEMA, which is our Georgia Emergency Management Association, which really agency, which also has very close ties. So those kind of people that just have a higher perspective, they can look over the horizon farther than us. But the really the biggest concern we have is if it's so good, you may not even know [00:32:00] that you've been faked out. So that is really the challenge. What I will say though, going back to the election process and having a paper ballot, is you the voter, you look at, you go ahead and make your selections. Then you have your ballot, you can look at it, did I get everything right? And once everything is right, whether it's hand marked paper ballot or with a ballot marking device, then you put it on the scanner. And then the scanner goes ahead and records it, it drops into the box. That's your ballot right in there. And if the race is so close, they can pull them out, and they can do a 100 percent hand recount. If there's a QR code on it, it doesn't matter. You just read the human readable text. Who did they vote for president? Who did they vote for this race? Whatever race it is. And that's what you can do in about 97, 98 percent of all ballots in America. See, the funny thing is, America is actually very good at running elections. Some people say they do the whole election in France, hand counting. That's good for them, but guess what? They have one race. Now, they have five people, seven people that run it. [00:33:00] One race. Have you looked at how many different ballot styles we have in Georgia? I bet you Virginia's not too far. Sometimes for the May primary, we could have over 30, 000 different ballot styles. We have 2, 500 precincts. We have 180 state reps. We have 56 state senators. We have county commissioners with different districts. Then you have to figure out all those different ballot styles. Over 30, 000 different ballot styles. They have one in France. I think that's pretty easy to hand recount. So you think about that. So that's why it's very complicated, but it's decentralized. And that is the strength of the system. And every election director, every poll worker in America takes an oath. And I believe that when you take a note to the constitution of your state and the American constitution, you have a duty to honor it. And I think people do that. I know the people in the military do. And they actually risked their life. That could be a blank check they just wrote when they take their oath of service. And so really that's the high bar that we want to set for everyone that's involved in the election [00:34:00] process. And I've had confidence in Georgia and throughout the entire country that's what they're going to do this cycle. And they'll be breaking your heart perhaps if you're on that side of the aisle. The other one you'll be doing cartwheels and handstands. But our job is to do our job. I think we've shown in Georgia we'll do our job. I think other election directors have shown they'll do their job as well. Tammy Haddad: Jen? Jen Easterly: Just to pick up on a couple points about the decentralization, because I think it's really important. You you talk about the elections and how you run them, 159 counties and the various laws that are in place specific to Georgia. If you've seen one state's election, You've seen one state's election. There are different rules in place for every state. And that decentralization adds resilience to the process. But the rules may be different in terms of when the ballots are counted. And it just goes to the point that Kyle was also making about the importance of the media. Because the [00:35:00] media really has to be careful, particularly in a world of generative AI, not to amplify inaccurate information or things that our foreign adversaries are trying to do to stoke incitement and fear and chaos and panic. And so really important that the media Focuses on amplifying the voices of the real experts, state and local election officials, and not feel like they have to rush to a decision, particularly if ballots are still being counted. So I think that's really important. You talked a little bit about physical security issues. And we also work with election officials to do security at first entry assessments, looking for physical security of polling places. We advise on things like the security of mailing because we've seen fentanyl threats to mail. I just think and you alluded to this, Brad I just think election officials I wore a uniform for 21 years, I was in the Army, and whenever you [00:36:00] go in some place and you're wearing your uniform, everybody thanks you for your service, which is wonderful, but at the end of the day, it's election officials that we should thank for their service, because they have been dealing with so much over the past four years, and many of them just quit because they couldn't deal with the threats to them and their families. But at the end of the day, these are folks that are not getting a lot of money, and are doing it because it's the right thing to do because they take an oath. And when you go to the polls take the time to thank your folks who are working there, because they really are the front lines of defending democracy. I think that's incredibly important to, to recognize. To recognize. And then just back to the AI piece, I think it's also important to talk about the generative AI companies. At the Munich Security Forum this year over 20 of the generative AI companies that are building these capabilities, signed up for a tech to a tech accord. So that they can put rules and guardrails in place to prevent [00:37:00] deceptive AI from being used to manipulate our elections. And so we really need to rely on them to ensure that they're authenticating content that's being delivered with our capabilities and labeled in a way that you can tell it was generated by AI. And so you can't. have this fake content being manipulated without voters knowing. I think that's really important. What I get concerned about and we've talked about this before, we've seen four decades of technology that never prioritized security. It was always about cool features, speed to market, driving down cost, and frankly, it's left us in a world where the Internet is full of malware, software is full of vulnerabilities, social media is full of disinformation. And that's why we're very focused at CISA on working with generative AI companies to ensure that they're building capabilities that put security at the very top of [00:38:00] priorities, because these capabilities are more powerful. More unpredictable, moving faster than any technology we've ever had, and frankly, we need to ensure that what is being created is safe and secure for use, not just for elections, but for all of our critical infrastructure that we rely on. Tammy Haddad: What's interesting, because Sam Altman and OpenAI, they're the only ones that have really come forward and said that they're going to pull anything that's anytime their material will be abused. But Kyle, to you the social media, so it sounds like you guys have the infrastructure to keep AI out of politics, for lack of a better way to say it, but then we have social media, right? And the socials, Have basically said we're not going to step into politics. We're not going to moderate. So Kyle how do you think that's going to work? Kyle Kondik: Probably not. A friend of mine is in the fact checking industry and I was talking to her recently and she was lamenting that they used to just basically get a lot [00:39:00] more business from the social media companies because social media companies were more concerned about fact checking and accuracy, and they've decided to maybe take a little bit of a step back and because the difficulty of any sort of content moderation is that either do none or you do a lot and it can be easy to just say I'm just going to get out of this business and it's I think it's just unfortunate that's happening. And the other just thing about the media environment that we're in today is, of course, that every single person is a publisher, basically, and every single person needs to be a sophisticated news consumer, much more sophisticated than they used to be, and even for people who are professional news consumers, it can still be difficult to to sift through everything and I do think it seems like There probably needs to be some more federal regulation of this stuff, although you think about like the Federal Election Commission and its role and unfortunately the FEC can, it's pretty easy to get around the [00:40:00] FEC or it's pretty easy to make the FEC deadlock and not really enforce things on advertising and whatnot. The other thing is that if there's some sort of. malicious behavior going on. One example is there was this bogus robocall in New Hampshire right before the New Hampshire primary. Of course, you remember Joe Biden was not actually on the ballot there. But it was this AI robocall that was supposedly Biden's voice, and he was basically telling people not to vote. After the fact, the authorities figured out who it was. But if you do something like that, a dirty trick right before an election the authorities aren't necessarily gonna be able to catch up and you're not going to be able to turn back the clock on that actually getting out there. And so that's a that to me is a big challenge here, too. Tammy Haddad: So please get your questions ready. We're gonna come to the audience soon, and Glenn has a microphone somewhere if you just very there you go. Okay, right there. But now we talked about all the negative scary things, and now we're gonna talk about the positive things and how you think if you do think that AI is [00:41:00] beneficial in elections. And how we live our lives. Brad, do you want to go first? Brad Raffensperger: I haven't seen the benefit yet. But at the end of the day, I was in the business world for 30 some odd years. Our business is just talking to people one on one and that's how we did business. And I think really our society needs to get back to talking to people, having conversations human being to human being. And when we have disagreements on policies to be able to bring in someone Whatever their political stripe is, as long as they speak respectfully to the audience likewise should be respectful to them. And then they can leave saying, I never thought about it that way. No, I'm still going to do it this way, but at least you've been respectful. And I think that's where we really need to get back to. I think I think that our founders would be disappointed in us, and I'm talking about Jefferson, what better place to talk about his name than here, but I think that they would be somewhat disappointed that we don't have more of these conversations. Now, I know that [00:42:00] Jefferson and Adams, they had their moments, but they finally made amends probably right at the end, but I think that we need to get back to just talking to each other. We're Americans. We're part of a big family. Tammy Haddad: One of the things I like about AI, and I started this Washington AI Network, is because it's like Google search when it started. You have more access to information immediately and curated specifically for what you're interested in. But just like anything else, you have to be careful. But Jen, you're the expert. What's the positive side? Jen Easterly: In terms of elections I'm not sure there's a lot of positives in the near term. I was just talking to my husband about this before I came up here. From a cyber security perspective, we think that there will be benefits in using AI to detect cyber threats faster, so that will be a good thing. But I continue to focus more on the threats and the risks because, quite frankly, these are the most powerful [00:43:00] technologies of our century, maybe of all time, and as opposed to the most powerful technologies and weapons of the last century. Nuclear weapons that were built and safeguarded by governments. These are all built by the private sector largely who are incentivized to create profit for their shareholders. So the incentives are different in terms of how quickly these things are moving and how secure they are. And as we know, there's never been technology regulation on anything. Again, that's why malware, software the internet. And there are about 34 states who have 50 separate bills around AI and putting guardrails around them. The federal government has been talking a lot about that. I hope the Congress moves forward and puts something in place so we don't just leave it to the states to do this. But at the end of the day, I think we all really need to be focused on what we can do from a [00:44:00] security perspective, even as AI, of course, will do amazing things to cure diseases and to provide education and tutors and all of this. But at the end of the day, our job is to really ensure that we are securing these very powerful capabilities from misuse. Tammy Haddad: I can see if you could take a, name a candidate, right? And you took all the information about them, everything they've ever written, right? And put it in one of the AIs, right? And out would pop this person who would, You would ask him any question. What do you think about immigration? What do you think about this? It would all come out. I don't think that's necessarily a bad thing. That's not someone you vote for, but you could argue that it would be another way to your point to dis in from I'm looking at you, Brad, because I think that a Brad AI would be good, although you've already gone to every single county in Georgia. All right, who is the first question? Jen Easterly: Brad GPT. Tammy Haddad: Brad GPT, yes. I [00:45:00] feel like you're like that now because you've traveled so much. Brad Raffensperger: Thank you so much. Tammy Haddad: Here we go. Right here, this gentleman. If you don't mind standing and saying your name, it would be great. Grant Tate: Hi, I'm Grant Tate. I'm a business person, consultant. One of the ways that we're using AI is for countering strategies and position papers. Given I've written a strategy, and then we use AI to say, please analyze this strategy, tell me the vulnerabilities, and how I would compete with it. And so it strikes me, the same thing, so my question is, how are we using AI to look at our own vulnerabilities? Tammy Haddad: Jen? Jen Easterly: Yeah, so we absolutely are. As part of the, thanks for the question. As part of the president's AI executive order that was published last November, we were specifically asked to run a pilot. It's more from a technical perspective, but to look for vulnerabilities on infrastructure so that we can get [00:46:00] ahead of threats, but just as we can look at it from a technical vulnerabilities perspective, we can look for vulnerabilities Using that across all sorts of critical infrastructure from both the cyber perspective, but also from a physical perspective and then from a documentation as well. So just to be clear there is enormous promise in these capabilities and they are incredibly exciting. Generative AI is captured the world's imagination and rightfully but I really believe that it is the job of leaders to be able to leverage the power of imagination, but also avoid the failure of imagination. And there is a lot of failure of imagination to be worried about here. And but of course there'll be some good. Tammy Haddad: What's so interesting that even the AI experts are all surprised by what the other company came up with. You think the U. S. will still be ahead on AI, right? You're not worried about China or anyone else creeping up [00:47:00] really fast in the next year or so. Jen Easterly: No, I'm really not. And folks say we shouldn't regulate AI because it'll put us behind. And I think that is just a false argument. China is actually heavily regulating their AI. At the end of the day, we need to ensure that we have the right guardrails to protect these capabilities from being abused by nefarious actors. But that doesn't mean that we can, that it'll prevent innovation. There is such a thing as responsible innovation. And there is no country more innovative in the world than the United States of America. And nobody should bet against America when we're talking about innovation. But at the end of the day, we cannot, as we have for technology for 40 years, just allow these capabilities to be built without any security. Tammy Haddad: Are there any regulators here? You need to get to work. Okay, who's next? Who has a question? Right behind you. You would stand up, please. Thanks. Robert Somisky: Yeah, I'm Robert Sinitsky. I wanted to [00:48:00] ask what kind of guardrails are there for AI taking information from where it grabs the information to formulate its projections or its output? And I'm thinking about social media, how there's so much input out there that could be drawn by AI that's inaccurate, that's false. Or generate it as being the truth when in fact it's not. So what kind of guardrails are there to help ensure what we're getting out of AI is really factual? Tammy Haddad: We don't know, right? Jen Easterly: There are no official guardrails. All of the controls that are put in place are put in place voluntarily. Many of these firms made commitments during meetings at the White House. Many of them made them as part of this tech accord at Munich. But there is no regulatory no rules in place to say you must put these guardrails, either in terms of how you train the capabilities, in terms of how you deliver those capabilities, in terms of [00:49:00] the security you put around those capabilities. So you bring up a really great point if you have garbage in you will likely have garbage out. And so these capabilities do something called hallucinate, which they can, you can ask them a question and they can give you information. That sounds very accurate, and it can be completely wrong. I like to call them frequently wrong, never in doubt. And there's a real worry about these hallucinations when you're asking questions. It's probably one of the reasons why Universities don't all use them because it could be inaccurate information coming out. And again, we really need to focus on how we can put rules in place to prevent nefarious actors from using this, particularly in deceptive ways. Now, the E U just put in place their own set of regulations, the EU AI Act, which basically tiers the uses of AI. [00:50:00] So if you're using it for medical purposes you want to have the most stringent, guardrails put around those capabilities rather than just It's using them for something that doesn't matter that much. And a lot of people criticize Europe for being very regulatory heavy, but I think in this particular case, it's actually important to, to look at what the Europeans are doing and to learn from the types of things that they're putting in place that we can actually incorporate into the legislation that hopefully the Congress will put together in the coming year. Tammy Haddad: And really everyone here should if you have concerns, that's why you call your legislators. It matters. I write down the amount of people that call. Anyone that tells you any different is not there, at least in Washington. It makes a really big difference. Okay, who has the next question? Tammy Haddad: There we go, this young man over here. Wyatt - Intern: Thank you all so much for coming. I'm Wyatt. I'm a second year intern at the center. You just talked about the EU AI Act. So I was wondering [00:51:00] what the legislative outlook is in the U. S. and whether you think bipartisanship around AI will hold for the foreseeable future. Tammy Haddad: It's the only bipartisan issue, really, right? Right now. But Do you want to jump in there? Brad Raffensperger: No, not about Washington. Tammy Haddad: Okay, good. Brad Raffensperger: I'm doing pretty good right now. Tammy Haddad: Yeah, I think you're in good shape. Jen. Jen Easterly: Cybersecurity for a long time and still mostly is a bipartisan issue. I am hopeful about it. Frankly, there have been a lot of hearings on this led by the majority leader himself a lot of good work done again in a bipartisan way. Whether they're actually able to get a bill through in this Congress. I wouldn't want to predict it with everything else going on. But I remain hopeful because. Tammy Haddad: Because that's who you are. Brad Raffensperger: Ever a diplomat. Tammy Haddad: Ever a diplomat, that's for sure. Kyle, do you want to jump in on this? Kyle Kondik: [00:52:00] Sometimes what happens when a new issue emerges, and this is true throughout history, is that it takes some time for it to get partisanized. And I wonder if maybe we'll get to a point where AI does get more partisanized, which probably would be a bad thing, but maybe now is the time to strike because it is a new issue and maybe the different sides are haven't quite figured out where their sides position on, on it may be. Now, of course, there's also just a, kind of a a populism against big tech that I think you could probably find on both sides. So I think it's. It's probably more prominent on the right at this particular point. And that will that'll play a role in how all this stuff evolves over time. Tammy Haddad: Politics, as always. All right, who's next? There we go. Jason Easterly: I'm Jason easterly. My wife is sitting up there. I'm a veteran and dedicated public servant. Secretaries Raffensperger and Jen I know that y'all were threatened, have been threatened, exposed to the risk of violence. And you're just two among thousands of officials across the [00:53:00] country. It's my thesis that the information environment has been so contaminated and that is really the driving force behind a lot of these threats. I know it didn't happen in this time, but you remember the man who went to Pizzagate, who came to D. C. with his M16 and was shocked to find that there was no child trafficking going on at the pizza shop. And I see A. I. as a force multiplier, if you will, for disinformation in that environment. What's the responsibility of elected officials, even though the information in the environment is largely controlled by private entities, social media, other corporations, what is the responsibility of public elected officials like the governor or the president or members of Congress to say this is not right in real [00:54:00] time to shut down these rumors because sticks and stones, right? But words actually can kill. Tammy Haddad: Brad. Brad Raffensperger: I think that elected officials at all levels have a responsibility. You're responsible. And so I think you really have to address that. But also then to, as it goes to like poll workers, people that don't have any type of security. I think that we really need to help them navigate through that. I think that doxing should be really treated as a crime. And so that is very dangerous. Like after the 2020 election, a lot of information got out there, whether I could actually scrub that out of my profile, I probably not, but so you can have a clean profile, but also just taking, posting photos. Has all your metadata there. People don't think about that. My wife's a photographer, so she knows all about metadata, so it's out there exactly where that photo was taken and then taking a picture with a window somewhere in your house or around your property. All those things to make people aware of what these threats could be [00:55:00] from. Is that what the situation I went? No, I'd like to go back to Mayberry, but that's where we are right now. So it's trying to help people understand what those threats could become from. I think it's a responsibility. Anyone that holds elected office. To be responsible and to call people out. It's not about political party, it's just to make sure that people understand you cannot threaten someone else and you need to make sure that the information is accurate without making it so personal. Particularly election workers, they're just doing their job. You think about what they get paid. Yes, they get paid more than jury duty, but you're talking about $150 a day. And so something like that, maybe it's a bit more in other places. It's public service. That is probably a 14 to 16 hour day on election day. It's just giving back to their community. Those are the people at the grocery store, the soccer field, football field. Good people, just like you. And they need to be supported. Tammy Haddad: Were you ever afraid? Brad Raffensperger: I was very cautious and careful. And it is what it was. Tammy Haddad: What about your [00:56:00] family? Brad Raffensperger: When your daughter in law, who's a widow, is threatened and people are tracking her through a store and with their videos and all that, fortunately, she did the same thing to them, so we know who they are, and we have their picture, and that was shared with law enforcement. And when we were, had people show up at her house we knew what states they were from. They were from Georgia. Go figure. They're several hours away. You make sure that you understand where we are right now, and you be as cautious as you can. But at the end of the day, we have a job to do, and I think what I've shown is I'm going to do my job. It's a responsibility to do my job. Tammy Haddad: Do we have any more questions? here we go right up here, two more, James Weissman: Thanks. My name is James Weisman. So I think you did a really good job of laying out the [00:57:00] infrastructure that voting is secure. And AI might twist that a little, the results, but I'm concerned that there's a second part. There's collecting the votes, then apparently with like false electors, what happens on the back end? And I'm just curious. If there's any new structures that have been put in place to secure that part. Brad Raffensperger: You're really talking about chain of custody. At each precinct that you have virtually across the entire country, but in Georgia, when the ballots are all collected at the precinct, first of all, they'll post the results on the door there so you can see exactly how the candidates did. Then that goes to the county and then the county all the ballots that are in the storage boxes goes in Georgia with law enforcement. So there's really a tandem there. to make sure it does get to the county. So there's a chain of custody and then you fill out the form what you got when you got it from what precinct so that you can follow that chain and those results that the county receives, then they collate that, [00:58:00] verify any inaccuracies, report that to the state level. So we have chain of custody throughout the entire process. I think that goes to answer the first part of your question, but the thing is that when you show up to vote, we actually record who you are. Then we actually record who actually touched the machine. We use ballot marking devices. So let's say 1, 000 people come in, we check off 1, 000 names. Then we check with the ballot marking device that there was actually 1, 000 ballots that were printed at this location. Then we actually look at the ballot box. You've got a thousand pieces of paper, whatever that number is, that all those numbers are lining up. And then we see what the county says they received. Oh, they got a thousand ballots. Check it off. Then we do a 100 percent hand recount if we have to. But typically we'll run those through the scanners again, and we're using a 95 percent risk limit, which sounds like there's a 5 percent plus or minus error. No, anyone that has done statistics, it's actually very, it's less than 1 percent difference delta based on the sample size you've actually picked. [00:59:00] And so I say all that is you're cued to actually do a 100 percent hand recount. And we also allow both political parties and any independent observers to come watch the process. And we think that's really important. That's very transparent. The end of the day, I can't do anything about dishonesty of other folks. All I can do is be honest with what the election process is and what I say to the public. And I want to make sure I do that in a calm, rational voice. So I've left my contractor language that left in the job trailer. Now I'm not a contractor. I'm Brad, the engineer. I'm Brad, the secretary of state. And so I'm going to give people the calm response when everyone else is losing their heads. I think Rudyard Kipling said it best that it's for me to actually be the, if in that poem, then I'll be the one that's a calm voice of reason. I think that's what America needs up and down the line from County surveyor, To State House, to the highest offices [01:00:00] we have in this country. And America's begging for new leadership up and down the line. For people that are principled. And we're going to get it someday. Hopefully soon. Tim Lambert: Hi, I'm Jim Lambert. I'm a member of the faculty here, and I'm wondering, could you say more about the workforce that's doing its best to count the votes and run the polls and all the rest, but they go home at the end of the day. They have access to open AI and Gemini, and they can type any issue in there, make queries. I'm pretty sure I signed or made a checkbox that anything I've typed into that, those prompts is open or property of the company. Is, are these workers, they're able to use it and then suppose that becomes public information? I'm wondering are there rules that should be in place or protocols or best practices for the, for that workforce to use or not use [01:01:00] AI for whatever reason during the election. Thank you. Brad Raffensperger: That's a good question. Really right now people are doing their job and they go home and we'll have to consider something like that. But also to that point, if you're really concerned about disinformation, we actually encourage people that are activists on both sides. You don't trust the process. Okay, we're going to give you poll worker training. We put in that into state law. So you have poll worker training, but you're just an observer. Actually do the really hard work show up for that 16 hour day and go through there and actually become a poll manager. But you're going to see all the checks and fail safes that we have in place. And that way you're going to say, this is really, Good process that we had here. And then you have to understand this is what we're doing at all 2, 500 precincts in Georgia. And I don't know how many precincts you have in Virginia, but they're going to follow those same procedures because that's the job. We have procedures, we have state law, and everyone is supposed to follow those procedures. Will there be honest mistakes? People do make mistakes, [01:02:00] but it's such minuscule errors. But when your state gets really close and competitive, we have to squeeze out any source of error. Because for a statehouse seat, six votes could actually make a difference. That's the kind of things we're trying to squeeze out. A presidential race out there in Arizona was about 10, 000 vote difference. You got to make sure that you really squeeze out any kind of inaccuracy you could have so that people have high confidence in the results. But we can't control what outside forces say. We understand that. And some people have lots of Twitter followers or X followers and other social platforms that are out there. Tammy Haddad: I think that's the other thing. You have to be skeptical, more skeptical on social media that people immediately buy everything they see, even very sophisticated people. If you see it twice you believe it. Remember the whole Hillary is a lizard in 2016? And now like 50 percent of the stuff out there just really doesn't make any sense. But yet you look at it. Jen Easterly: I think it [01:03:00] goes to the individual responsibility not to, and because social media is made for engagement and what engages, what causes people to engage. More than enragement. So a lot of this is out there to get people to amplify it and to create that partisan discord, which, of course, our foreign adversaries are looking to do to undermine confidence in the fundamentals of our democracy. Everybody should be a critical thinker in terms of what they see. And if it doesn't sound right. Don't retweet it, don't amplify it, ask questions around it. And again, to Brad's point, if you have any questions about the security or integrity of elections, be a part of that process and go talk to those folks who are the authoritative sources and the subject matter experts, so you can help us protect our democracy. Tammy Haddad: That's right because it's everyone's job. I remember the early days of those fake news sites and the [01:04:00] names of the Reporters all had like 42 consonants in it, so you knew that it was Russia because it was like, whoa no one has a name like this, but now it's so smooth. This has nothing to do with AI, but it's so smooth and it's integrated in, which is why you should all support media. And you guys need to start focusing on media. Thank you guys so much. Tammy Haddad: Thank you for listening to the Washington AI Network podcast. Be sure to subscribe and join the conversation. The Washington AI Network is a bipartisan forum bringing together the top leaders and industry experts to discuss the biggest opportunities and the greatest challenges around AI. The Washington AI Network podcast is produced and recorded by Haddad Media. Thanks for listening. ###