AI Deepfakes, The Cornell 7 Mystery & Life With an Astronaut

America At Night with McGraw Milhaven

The episode opens with UCLA Professor Ramesh Srinivasan discussing the escalating threat of AI-generated political deepfakes and the urgent need for federal transparency laws to protect democratic processes. Host McGraw Milhaven then brings in

Key takeaways

• UCLA Professor Ramesh Srinivasan warns that AI can now create hyper-personalized political ads targeting individual psychology rather than broad demographics.
• California and the EU have enacted data privacy regulations, but the U.S. lacks federal laws mandating disclosure for AI-generated campaign content.

Main topics

• AI deepfakes and political advertising regulation
• Cornell 7 sexual assault allegations and legal proceedings

Notable quotes

"If nobody knows what's real and no one trusts anything or anybody, yet we're all addicted to technologies that we know don't necessarily tell us the truth, what are we going to do?"

Conclusion

The episode balances urgent technological and legal concerns with personal

Transcript preview

Speaker 4 (0:05) Westwood One presents America at Night. Here's your host, McGraw-Milhaven. Let's Speaker 2 (0:12) get right to it. Thanks for joining us. Lots to get to. We'll talk about the Cornell 7 a little bit later on. There's so much misinformation about that. We'll get to all that. Also, Matt Wolfe, talk a little entertainment. But let's start off this first hour. Welcome, everybody, America at Night. It's Coast to Coast with McGraw-Milhaven. So apparently we are in election season, and artificial intelligent commercials, videos, are now in campaign ads, and people have a hard time trying to figure out what is real and what is fake. We're going to talk to a man who knows a thing or two about this. He is the host of the Utopias podcast. He's also a professor of information and studies at UCLA. Professor Ramesh Srinivasan joins us. Professor, welcome to America at Night. Thanks for joining us. Speaker 1 (1:03) Hey, it's great to join you guys. And it's a really cool show that you're running. So I'm really happy to reach the wide audience that you have. Thank you very much. Important issues. Yeah. Flattery Speaker 2 (1:14) will get you everywhere on this show. Everywhere. Everywhere. Speaker 1 (1:20) That's Speaker 2 (1:20) what Speaker 1 (1:20) we like as professors. Just kidding. You got it. Speaker 2 (1:23) Let's talk about AI-generated commercials. How big of a problem is this, Professor? Speaker 1 (1:29) It's a massive problem. Look, for the last, I would say, 10-plus years, increasingly, I think all of us know that it's become harder and harder for us to trust anything we see online, right? So, you know, if you just roll the clock back 10 years or so, so many of us were deeply concerned and distressed by these examples like Cambridge Analytica. and other third parties that basically manipulated what we see online. Deep fakes have been a longstanding issue. But imagine if there's a technology that is monitoring you, probably constantly, in ways that you wouldn't even know, that is constantly gathering data about you, what you do, what you click, your voice, your face, etc., and can essentially use all the data gathered about you to imitate, simulate, or replace you. And that is what we are staring at, which is so, so, so critical and so much of a problem that we must overcome, especially with the midterm elections coming, but in general to protect our democracy. Because if nobody knows what's real and no one trusts anything or anybody, yet we're all addicted to technologies that we know don't necessarily tell us the truth, what are we going to do? Speaker 2 (2:44) Yeah. Are there any, let's first start with any regulations. Are there any rules or any regulations when it comes to advertising and artificial intelligence? Speaker 1 (2:56) Nothing significant, certainly not on a federal level. On the state level, we have some here in the state of California and other states, there are some regulations around data privacy. The European Union has some pretty significant regulations around privacy and data. They even have what's called an AI Act that is pretty good in lots of ways. But as you probably know, people in Europe really feel like... they are like outside of the game when it comes to AI and developing AI. And they've felt that for a long time when it comes to technology itself. So, you know, what we have to do in our country is support our entrepreneurialism and our innovation, which is not exactly what's happening with AI, because we're only talking about two companies really at the end of the day that are dominating market cap and are likely both to go public, Anthropic and OpenAI. But what we really need to do is, sure, like continue the innovation, but we have to do, we have to really rethink what innovation is. So we ensure that every single stakeholder in our country is protected. You know, so we're protected as workers, we're protected as consumers, we're protected as citizens and a democracy. These are critical identities that must be protected and AI should not be violating those. Speaker 2 (4:10) Professor Ramesh Srinivasan is our guest. He hosts a podcast called Utopias Podcast. He's also a professor at UCLA. And if he sounds or looks familiar, you might have seen him in the new Netflix movie, The AI Doc. So, Professor, we appreciate you coming on. We're not just talking social media. Are we talking TV commercials, cable commercials, radio commercials? Are we talking that? Are they at that level now? Speaker 1 (4:43) Well, it really depends on what protections broadcast media have to try to ban AI-generated content. But I personally believe that any single piece of content, video, audio, photo, et cetera, text, if it is AI-generated or significantly AI-modified, I believe all of that should be watermarked. So I think it's up to the networks, the broadcast networks, the cable networks, radio networks like yourselves. to actually have, you know, and at some point there were FCC protections around these issues. But right now we're just sort of seeing this administration completely bend the knee on every single level possible to these AI companies, even though they're wildly unpopular because nobody sees how those companies and what they're doing is going to benefit them. They just see it as further division. So it's really up to... There is a lot here. Yeah, please. Speaker 2 (5:38) There is a law that says... You know, you see the laws are so nebulous and so confusing, right? Sometimes the candidate has to come on and say, you know, my name is John Smith and I approve this message. Others are our third party PACs, which don't have to disclose much, if at anything. Is there any law that says they have to disclose anywhere in the ad, even in the fine print? This is this is AI generated. Speaker 1 (6:02) Not specifically and not as of yet, though, that is absolutely the bare minimum of what we need. And we actually really need it on a federal level. The very bare minimum that is needed is transparency and disclosure. But quite honestly, what we need is more like accountability. And we really need different stakeholders that are affected by these AI technologies to have seats at the table to essentially... audit these technologies and make sure these technologies are going to benefit the groups that they represent as well. Because remember, all of these technologies are only worth a penny because they learn from our data. But imagine if you have like an AI that's constantly monitoring you and it's going to basically lead us to the place where it's no longer going to be, you know, you having to do this hosting and broadcasting of America at Night. but it's an AI that replaces you. And this is where we're heading more and more. Speaker 2 (7:00) So is AI doing this on its own? Is there somebody directing it and then it's sort of created on its own? Or is this human-led where the humans are going in and creating the ads and then telling AI to generate it, if that makes any sense? Speaker 1 (7:18) Yeah, that's a good question. I mean, at the end of the day, AI systems and technologies, as we're starting to see more and more, are taking a life. on, if you will, of their own. And that is because they are instructed to achieve certain objectives and they are going to do whatever it takes to achieve those objectives and or communicate with other AIs that might have other objectives or other behaviors and then take on a behavior that cannot be predicted. But at the same time, most technologies follow the will of their creators. The problem is that we're seeing now AI technologies further and further diverge from the will of their creators. In this particular case, if we're talking about the creators being these LLM AI companies, I think we can see quite clearly that they don't have a clear sort of set of ideas about what they want the United States to look like economically, politically, and so on. And that's not really something I necessarily blame them for. That's up to all of us in the public sector. That's up to our government to not be sold out to big tech, which we've seen throughout the Trump administration. bending the knee from the inauguration day one. There are umpteen numbers of examples of the coziness between big tech and this administration, which means that the administration and our Congress, as we've all seen, has no capacity or power or maybe even any interest to regulate these companies, not to kill their products, but to ensure their products benefit multiple stakeholders. Speaker 2 (8:47) There was an article in Axios last day or two that was talking about that some campaigns are starting to get served with legal notices, cease and desist, you know, take down. So, I mean, clearly they are out there. Obviously, some are parodies, some are not, but we're starting to see at least some of these campaigns fight back. Speaker 1 (9:12) Yeah, which is good. And this is why data centers and AI, which are really one in the same, you know, in a sense, they're really all connected to each other, are such a huge issue right now that all of us are staring at. Because, I mean, check this out, McGraw. So like, for example, more and more, the way AI is going to function is you will be, let's say you're on Instagram or whatever, and AI will be targeting you. And I don't mean that in the inherently intense way the word target means, but essentially it will be personalizing its content for you based on you and only you, not your wife, not your kids, not your friends, not people of the same political party, not your neighbors, not the wider country. It's going to be based on what is known about McGraw, right? What is known about McGraw can be turned into data that is collected about you. And all of this data can be correlated, which means pattern recognition, with predictions of what will influence McGraw's behavior in this way, that way, or the other. And that is a very specific thing. That's not captured by demographics like age, gender, and so on. It's captured by the specifics of your own psychology tied to data being gathered about you and only you to influence what you see, which is then going to influence what you think and how you feel. Speaker 2 (10:30) So are you saying that in the very near future, these ads are going to be created? One's going to be created for you and another one's going to be created for me and another one's going to be created for my 19 year old daughter. Speaker 1 (10:44) Exactly. Yeah. And those might be pretty different ads. So you all might be say the ads are all about, let's say, President Trump or, you know, I don't know, you know, Bernie Sanders or what have you. They could be completely different ads from one another. And we. not call them ads, we might just see them as quote-unquote content or stories, right? So that is why, and pretty much all of us feel intuitively like that this is kind of what's happening. That is why Americans, and Pew studies have shown this again and again, do not trust what they see online. But I think we all know that... our phones have basically become almost parts of our body. Like, we can't really go without them. Well, I mean, Speaker 2 (11:25) look, we've seen Facebook as one giant cesspool of either pornographic ads or deep fakes or whatever else. So, I mean, interspersed, there's a commercial for a new type of pen, but otherwise it's all garbage and it all just sort of follows you around the Internet. I don't really have much hope because a couple years ago, They banned robocalls, right, because it was annoying and no call lists. And you were sitting down for dinner and your phone would ring every night. But what the elected officials did was they exempted themselves. So they themselves were allowed to continue to robocall, but all the other businesses had to stop. Speaker 1 (12:06) I mean, back in the day, I know this is 20 plus years ago, all of those companies were on the verge of being regulated, so they couldn't gather data without our consent, where the opting out of warrantless data collection would have been what occurred. But that was when 9-11 happened and things like the Patriot Act got shoved down people's throats without us fully understanding it. Patriot Act was supposed to be about protecting us from terror, but it ended up being basically opened up for data collection writ large. And that's why we saw, you know, some years later, the Snowden revelations and other examples like that. So this issue is this issue McGraw is not unique to Meta and Facebook. I mean, these are companies that are in the interest of capturing our attention, influencing our behavior. And AI technologies have the capacity to read, write, speak and think for us, which makes us all wonder, why do we matter and how should we matter? And I think it's time more than ever for everybody, and these AI companies should be paying for it. Not only should they be covering all their energy expenses and bringing the prices down, they should be paying us for our data, but also they should be paying for really true literacy programs that give every American the capacity to understand how these technologies work and how these technologies can benefit them and their own creativity. Because it is entirely possible for these AI technologies to be tools. I mean, remember, technology is a tool. It can be a tool to actually expand what we do and how we do it. And that is very cool. But that's the totally opposite. That's the augmentation of human intellect and creativity, not the replacement. Speaker 2 (13:43) Professor, I got one minute left. You know, the opposite is also true. If they can create lies about you and make them real ads or real voices in radio commercials, well, then. They could create a true ad about an elected official and the elected official can say, oh, that's only fake. So you don't know. Right. You don't know what's real and what's true. And you could say what's true is fake and what's fake is true. And no one knows what's going on. Speaker 1 (14:11) And we're all just going to go cray cray. Right. So that's why that's why we can do it. What is it? We can we can enact. Right here and now, because it's wildly popular, forget these sort of so-called partisan divisions in our country, Republican, Democrat, left, right, whatever it is. People across the board want AI technologies to not take us off the rails, to not increase inequality, to support the affordability crisis. And it's entirely possible for us to regulate, which is basically direct a car that might be going out of control. The car is not destroyed. The car just is going on the right road. And that is what we need to do. And I truly believe we can do it. And there are a number of really good ideas out there. It's a matter of achieving the political will. But for all of us as Americans to know on our own that we should not fully trust these technologies. Let's understand how these technologies work on a basic level and use them to benefit us. But do not see them as black mirrors or just places that we dive our head into and just blindly trust. Because that's going to divide us and polarize us even more. Speaker 2 (15:13) That is Professor Ramesh Srinivasan. Close enough, right? Srinivasan? All right. The podcast is Utopias. He's also on the new Netflix movie called The AI Doc, and he's Speaker 1 (15:28) a Speaker 2 (15:28) professor at UCLA. Professor, thanks for checking in. Speaker 1 (15:31) Really enjoyed it. Thank you for having Speaker 2 (15:33) me. Stay right there. We're just getting started. America at Night, back in a moment. Speaker 4 (15:39) America at Night with McGraw-Milhaven, where the nation comes to talk. Speaker 4 (15:56) This is America at Night with McGraw-Milhaven. Speaker 2 (16:00) All right, we're just getting started. Next hour, we're going to talk to Ed Mitchell. Ed Mitchell was on Apollo 14. He has since passed away. But that was the Apollo mission that landed on the moon where Alan Shepard hit the golf balls on the moon. The other person, he was the other guy, Ed Mitchell. And his wife wrote a book. And she's got great stories about being the wife of an astronaut, of a man who walked on the... Surface of the moon. So we're going to talk to her next hour. Matt Wolf's going to be along to talk some entertainment. Lots of things going on in the world of entertainment. We've got to track him down. And then stay right there because this Cornell story has burst onto the scene, and there's a lot of misinformation, half-truths, conflicting stories, what was said, what wasn't said. It's very confusing. And before you have an opinion on it, I would prefer you actually have the facts. And so we're going to talk to Rich McHugh, one of our go-to at News Nation, who's been covering this story. And we want to get the facts. You can have your own opinion, but you can't have your own facts. And so we're going to check in with Rich from News Nation to find out exactly what is true, what isn't true, and what do we know and what don't we know. And that's coming up next. Stay right there. America at Night, Coast to Coast, with I'm Ron Millhaven. Back in a moment. Speaker 4 (17:23) You're listening to America at Night with McGraw-Milhaven. Speaker 4 (18:16) You're listening to America at Night with McGraw-Milhaven. Speaker 2 (18:20) Well, this week, this Cornell gang rape story burst onto the scene. And ever since it did earlier in the week, I think we've been trying to get our heads around exactly what we know, what we don't know. Our guest, News Nation investigative correspondent Rich McHugh. We literally took him out of the TV studio. He's got to get back in a second. We appreciate him taking time out for us. Rich McHugh, welcome back to America at Night. Hey, good to be here, always. You betcha. So this alleged crime happened in October of 2024, but it burst onto the scene this week. She filed a lawsuit, right? That's why this burst onto the scene this week? Speaker 5 (19:00) Yeah, that's right. About two weeks ago, filed a lawsuit, and it exploded into the national conversation because in the lawsuit, you know, there's seven defendants that are named. There's also Cornelis named. The bar is named. And so instantly the entire world knows the names of all these people. The allegations, you know, it's very the allegations are shocking. And so I think, you know, that's why you're seeing this outrage right now across the country. Speaker 2 (19:33) Yeah. And the allegation is what? What is she alleging? Speaker 5 (19:39) Well. that she was sexually assaulted by seven members of this fraternity, that she was plied with drugs, you know, ultimately that it was against her consent. And, you know, it always comes down to consent, right? Did you consent to sexual relations? And that's the whole crux of this argument, you know, this whole case is she says it was not consensual. And I think that's going to be scrutinized because there's a, there was drinking involved, lots of drinking. And I'm not, I'm not going into this to, to, to defend anybody in this situation. I'm just laying out why, why it's complex. Speaker 2 (20:26) Sure. Speaker 5 (20:26) Everybody involved was drinking intoxicated. There was also ketamine, this drug ketamine, which they were all doing. And in text messages, that she reportedly sent in the days following, she said something to the effect of, like, you know, none of the sexual stuff was illegal. You know, that was to one of the people she later accused. In a conversation with Cornell police. Weeks later, she reportedly told them, you know, this was 100 % rape. And, you know, that's the discrepancy on her side. If you read the text messages that are included or the snap messages that are included in the lawsuit. Coming from the fraternity members, they're just egregious. It's just, it's horrific. It's disgusting. You're kind of like, I can't believe that nobody stopped them, stood up and stopped and said, wait, what the heck is going on here? You know, at one point they're texting and they're saying free, blank, downstairs. And then later someone writes, you know, is shop still open for business? And nobody's chiming in to say, hey, guys, like, stop. And so that shows a lot. You know, it shows just at the very least incredibly bad judgment and disgusting kind of morality on the other side, the boys, the men. Speaker 2 (22:01) Yeah. Speaker 2 (22:04) Yeah, no, morally reprehensible. Legally, that's to be determined. But, Rich, why is it that because the prosecutors are talking and the prosecutors are sort of, I don't want to say defending themselves, but trying to explain why they didn't charge any of these seven. What are the prosecutors saying about this? Speaker 5 (22:25) Well, the prosecutor said in a statement, he said Jane Doe's signed sworn statement from 2024 did not allege she was drugged against her will or gang raped. On the contrary, her statement said, described her participation in drug use and sexual conduct as voluntary, conscious, and consensual. So that is the prosecutor. who decided not to prosecute and decided not to investigate further, coming out now to say that's what she said in the sworn statements in 2024. And so there's all these strands that are clouding the factual timeline of this whole thing. And so, you know, what really is needed is just a real investigation, unbiased, that establishes a timeline and a fact base. And the evidence should lead this conversation instead of the outrage that I think everybody is, you know, it's okay to feel outraged by what you're reading. But I think, you know, ultimately there's a sense that they're trying to say these seven are 100 % guilty before, you know, facts are established. And let's say they are charged and it goes before a jury. Ultimately, this will be up to a jury if they're charged. And so Speaker 2 (23:52) that's Speaker 5 (23:54) what we hope for, right, is that evidence leads this conversation. Speaker 2 (23:59) Right. And I can, no question about it. And that's why I think you sort of, at this point, we're still gathering the facts. There's still a lot of things we don't know. What about, though, last night, Rich McHugh from News Nation, there was a story that in the police reports that she filed, I guess a few days later, that she apparently said something along the lines of, no, it was 100 % rape. And apparently those statements from her never made its way to the prosecutor. Have we followed that threat at all? Is there any truth to those allegations? Yeah, the Speaker 5 (24:37) prosecutor is saying that never happened. But apparently that claim of hers is included in a police... detailed interview, long interview, and the transcript of which I believe CBS New York has reported obtaining. I have not seen the transcript, so we're reporting from somebody else's reporting. But the prosecutor is saying, I never received that. So this is a game of who knew what when, right? And if she did report that and the prosecutor... did know that and did not like even interview her. Well, shame on him. Uh, should be interviewing somebody. But if he did not know that, and the sworn statement was given to him as part of like the investigative package. And it said, all of this was consensual and she signed it. You can see why he's putting out the statement right now saying, look like this is what was presented to me. And this is why, you know, Sexual assault in America, it's so hard to investigate and prosecute, and especially on a college campus. Like these Title IX investigations that happen on college campuses, they almost always fail because they're not people that are trained in investigating sexual assault. It's like having a corporate HR team investigate something like this, and it needs a much different set of expertise. Speaker 2 (26:12) It's a really interesting point. Cornell, however, did discipline these seven, did they not? Speaker 5 (26:18) They did, to varying degrees. Some were expelled. One is, at least one, I think, is still a student there. Some were, you know, were kicked off campus for a while. I don't know the true extent of what happened to them. You know, a point needs to be made here. We have not heard a lot of their defense, so that will come. And I'm sure, you know, between all the people that are named in this lawsuit, they're going to get their chance to say, their side of the story. And so I know all these things go, you know, I've done a lot of reporting on sexual assault and then you get your line lands in a courtroom and all of a sudden you hear from the other side and you know, a jury is going to have, if it goes, if it goes, you know, indictments happen, a jury is going to have a tough time with this, with this case because of the alcohol, because of the text, the communications after the fact. you know, people are admitting that they don't, it's a foggy, you know, what they remember, they don't totally remember. It's going to be, it's going to be a landmine. It's going to be very, very difficult. Speaker 2 (27:24) It was reported at some point. I know this was a social media report, but did Cornell. make any of the seven as punishment write an essay? Was there any truth to that? I Speaker 5 (27:40) believe that as part of it, I think that was over kind of hyped as like that was the punishment.