Loading summary
Transcript82 lines
- [00:02]
Erin Ryan
Over the same old news cycle, tune in to Hysteria, your weekly group chat with me, Erin Ryan, and my co host, Alyssa Mastromonico, where no topic is off limits. From politics to pop culture, we're bringing you brutally honest takes on the stories shaping our lives. From powerhouse women like Elisa Slotkin to wellness trends in education.
- [00:18]
Alyssa Mastromonico
No sugarcoating, no doom scrolling, just real talk, strong women and hope to keep moving forward. Catch Hysteria wherever you get your podcasts and tune into our YouTube channel for full episodes and our special series.
- [00:30]
Disinformation Expert / Journalist
It is becoming increasingly hard knowing what is AI and what isn't AI. I don't think we're cooked quite yet. We're kind of like the frog in the water and it's getting to boiling point, but it's not quite at boiling point yet. We still can jump out. I'm hoping we as a species will.
- [00:46]
Akilah Hughes
But I don't know what's worse than an illegal war in Iran, killing a building full of schoolgirls, throwing international trade into turmoil, and wearing a stupid baseball hat when American soldiers come home in boxes. Oh, it's that plus this proliferation of fake AI generated war images and propaganda. By now you've seen the video game footage, the fake strikes in Tel Aviv, the fake video of US soldiers captured by Iran, the mislabeled clips from movies and older conflicts, and a lot more. I'm Akilah Hughes, and on today's episode of How Is this Better? I wanted to figure out how a void of real information and a plethora of idiotic AI prompts might change war and the world more broadly forever.
- [01:38]
Interviewer / Host
How widespread is this? Do you have any idea of, like, the scale of how many users are putting out these images, how popular they are, how fast they're spreading?
- [01:46]
Disinformation Expert / Journalist
I think this conflict, for a few years now, we've been kind of hearing worries about, you know, the coming AI apocalypse in terms of disinformation. And I think this for over the last couple of weeks or 11 days, however long it's been happening. Now, this is the first time that I've noticed AI being a prominent or predominant part of the disinformation landscape. And also the fact that AI tools have become so sophisticated, it is becoming increasingly hard even for people who are skilled in, you know, knowing what is AI and what isn't AI from knowing what's real or not. There are millions and millions of people who are seeing this AI.
- [02:29]
Akilah Hughes
This is David Gilbert. He's a reporter for Wired who covers disinformation and online extre. His latest piece is titled Fake AI content about the Iran war is all over X. So who's making this content and why?
- [02:41]
Disinformation Expert / Journalist
There are hundreds, at least, if not thousands of accounts who are sharing this AI that's ranging from everyone from Iranian state officials and state media sharing AI images right down to blue check mark accounts on X who are subscribed to its premium service and therefore can earn money from images that go viral. So we're seeing a huge spread of people sharing it, we're seeing tens of millions of people consuming it. And it is very quickly becoming clear that this is now a key part of how disinformation is going to change the news landscape during breaking events like the Iranian conflict.
- [03:19]
Akilah Hughes
If hearing that Iranian state officials and state media accounts spreading disinformation by posting AI slop gave you a certain kind of pit in your stomach, I understand. But if that sounded like a foreign problem, one that you imagine when you think about state controlled propaganda machines like in Russia and North Korea, or you
- [03:37]
Interviewer / Host
might be forgetting President Trump shared a
- [03:39]
Evan Osnos
fake AI generated video on social media showing former President Barack Obama being arrested in the Oval Office.
- [03:47]
Erin Ryan
He posted an AI video on his truth social platform showing him flying a fighter jet dropping what appears to be feces on protesters.
- [03:56]
Interviewer / Host
The White House here in the US also has made a name for itself and sharing a lot of AI images
- [04:02]
Akilah Hughes
since Donald Trump returned to office. Do you think that that has any
- [04:05]
Interviewer / Host
impact in like, how believable they are to people? Like if it's coming from official sources in separate ways and now we have this war, is that impacting?
- [04:12]
Disinformation Expert / Journalist
Yeah, I'm not sure if it, if it goes to believability, but it definitely goes to normalizing the use of AI and the fact that people now have become just so attuned to seeing AI images and thinking that because the White House is producing them and is disseminating them, it gives it a sense of credibility that AI images, you know, you can use AI for propaganda and promotion. And so when people then see it during breaking news conflicts, it's not necessarily, I think that they may be believing it, but that they just take it as this is part of how states communicate with their followers during times of crisis. Yeah, like the White House's use of AI in a lot of cases, they're sharing stuff that has been found online. Like if you look at Donald Trump's true social feed, it's just filled with reposts. The most incredible AI images of him, you know, as a, as a warrior, as this jacked kind of all American guy.
- [05:11]
Interviewer / Host
Yeah.
- [05:12]
Disinformation Expert / Journalist
Young man, young Man, Yeah, with great hair. So that, that's one level of it. But I think the stuff that's coming out from the White House X account or Facebook account or Instagram account, that's much more slick. And I think it's. It's a way of messaging to their followers, getting the message across that they want. And it's. It's really dangerous because it just makes AI content so normalized now.
- [05:36]
Akilah Hughes
Okay, so we're obviously not buying the slop from Trump depicting himself as handsome, but you'd be forgiven if you were fooled by an AI generated video a time or two. The technology is rapidly advancing and creating moving images in a much more convincing way. I mean, you may recall that AI version of Will Smith eating spaghetti from just a few years ago. Fast forward to now, and I still think he's taken pretty big bites, but it's way more believable.
- [06:02]
Interviewer / Host
Can you tell when it's an AI image now? Now that it's getting this good, or are you often checking to see, you know, is this real, is this fabricated?
- [06:10]
Disinformation Expert / Journalist
Yeah, I'm completely checking. Like, Google has a tool that allows you to input an image, and it will tell you if it was created with Gemini AI. And that's what I use. I use the tools that are available. But even those tools. There was a report came out a couple of weeks ago showed that Google's own tool cannot tell every time when it's shown one of its own images. So it's just becoming incredibly, incredibly hard. And it is changing by the day, if not by the hour. I spoke recently to Hani Farid in Berkeley, who's one of the experts in image forensics, and he just said that it's. It is now getting to the point where he can't tell anymore, not only in images, but in video. And video is always the one that was kind of, you know, it was quite easy to see in video something that was created by AI. Now we've got to the point where, at least for the casual viewer, it's. It's really, really difficult to tell at first viewing something that is AI and something that isn't.
- [07:04]
Akilah Hughes
There's another wrinkle to all of this. On X, people have started using the platform's AI chatbot, Grok, as a kind of instant fact checker. If a video goes viral, users will tag Grok in the replies and ask, is this real? Where did this happen? And so forth. The idea is that AI can scan the Internet faster than a human and give an answer. But as David reported, that System has a pretty major flaw. I want to talk a little bit
- [07:30]
Interviewer / Host
about this added layer here, which is the example you used to open your most recent piece.
- [07:34]
Akilah Hughes
Can you tell us what happened when
- [07:36]
Interviewer / Host
the disinformation expert you talked with tried to use GROK to verify a post that showed missiles that had supposedly struck Tel Aviv?
- [07:45]
Disinformation Expert / Journalist
Yeah. So it was Talhagen who's a disinformation, open source intelligence researcher who has been tracking this information coming out of the Iranian conflict from the beginning. And I think he got to the point where he was just so fed up with people relying on Grok, AIs, X's AI chatbot to verify images and GROK getting all those responses wrong, and not all of them, but a lot of them wrong, that he. He said, okay, finally, one last challenge. Show me or tell me, Give me proof where this missile attack took place. And so GROK responded multiple times. GROK got the answers wrong repeatedly, could not provide actual evidence for the claims. And finally TAL gave up and gave the answer. And in response to that, GROK produced an AI image of its own. Wow. As evidence of its own claims about the veracity of the video. So we're just getting deeper and deeper. There's levels upon levels of AI responding with AI to questions about while information is real or not. And I think it just. The more and more you rely on Grok or any AI to verify information, I think while it is AI can be really, really good at certain aspects of that. This shows just how dangerous it can also be because in a lot of cases, when it's responding to people, it's reflecting their own biases, it's confirming things that it thinks they. The person who asked it wants to hear.
- [09:19]
Akilah Hughes
Yeah.
- [09:20]
Disinformation Expert / Journalist
It's not giving a clear and defined answer. So that that example is a really just terrifying escalation of how AI is making the problem of AI disinformation on X so much worse.
- [09:33]
Akilah Hughes
Worth pausing and just putting a very fine on what happened here. A researcher asked grok, an AI bot, to verify whether a viral missile strike video was real. GROK got the answer wrong, and then it generated its own AI image to try to prove it was right. In other words, an AI system used fake evidence to defend a false claim about another piece of fake AI generated content.
- [09:57]
Interviewer / Host
Do you think that this is like a hallucination, or do you think that this is like a programming error?
- [10:03]
Disinformation Expert / Journalist
I think even most people are unsure, especially if you are not inside xai, the company that created grok it. Is it very, very difficult to tell whether this is something that has been, you know, decided upon by someone in a boardroom that, you know, to provide evidence we should allow GROK to create its own AI images.
- [10:23]
Akilah Hughes
Yeah.
- [10:23]
Disinformation Expert / Journalist
Which if that did happen, is obviously a really disturbing and worrying development. Or it is just, as you say, a hallucination that the. The AI itself has decided to take upon itself to do this. But either way, the result, I think, is that people are once again less informed and have less trust in the information that they're seeing online.
- [10:46]
Akilah Hughes
When we come back, more on what this means for war coverage and the increasing challenge of deciphering fact from fiction.
- [10:55]
Brian Reed
I'm Brian Reed. When I created S Town, I looked at how secrets lies and the stories we tell shape a small rural town. Now on my podcast, question everything, I'm going bigger. Hi, this message is for Senator Lindsey Graham. I'm hoping I head to Washington to take on a law that gives tech
- [11:12]
Evan Osnos
companies sweeping immunity is how these companies have gotten rich.
- [11:18]
Brian Reed
Join me as I go after big tech on Question everything from placement theory and KCRW out Thursdays, wherever you get your podcasts.
- [11:26]
Akilah Hughes
It's not just that we're getting flooded with AI generated videos and images. Increasingly, the accounts reacting to online content may not be people either. Researchers are now warning about the rise of AI bot swarms, networks of automated accounts designed to post, reply and amplify content across platforms.
- [11:46]
Disinformation Expert / Journalist
I reported recently about this idea of AI bot swarms that researchers have looked into, which is we've seen bot networks in the past from Russia, but those have been pretty clunky and, you know, relied on humans to prompt them to what to say in each interaction. What these researchers did was looked at how AI could be used to create giant swarms of accounts on social media platforms like X. Each one of them would be, you know, unique, individual, create responses in real time, be able to judge how to respond in real time to people interacting with us. So you're very quickly getting to the point where you no longer think or know if you're speaking to a real person online. If you're speaking to an AI bot and then you've got an AI chatbot in the middle of that who is creating AI images. You're just quickly, very quickly going to the point where you no longer have trust in anything that is online and who is behind it. And therefore, the amount of content that is created by AI is very soon, if not already taking over the amount created by humans on these platforms.
- [12:56]
Interviewer / Host
That's horrifying.
- [13:00]
Akilah Hughes
And with AI not just generating content, but also Spreading it faster than ever. It's worth asking, has the technology finally arrived in the battlefield of public perception? Could this be the first conflict shaped at scale by AI?
- [13:13]
Interviewer / Host
I mean, is this the first AI war?
- [13:15]
Disinformation Expert / Journalist
I guess it's the first war or first conflict that I have seen where AI images have been deployed at scale that are relatively realistic, that are not clearly and easily identifiable as AI. There are clearly a lot of the networks and the accounts who are seeking to take advantage of, of the monetization aspect of these platforms have been working on how to use AI to boost their engagement because the way they are deploying it is clearly designed to get as many clicks as possible before these are taken down. X has said that it is demonetizing accounts that it finds that haven't self labeled their content as made with AI. But I've come across one piece of content that has the made with AI label so far during my investigation on this. Yeah, and it's, it's a new, you know, it's not there. They haven't really announced it yet, but I was speaking to someone from X yesterday and they said, yeah, it's there. People, you know, people can, can use it, but they're not using it.
- [14:20]
Interviewer / Host
Right, exactly.
- [14:21]
Disinformation Expert / Journalist
Yeah. I think this, this conflict is definitely, seems to be a tipping point in terms of how AI is being used and how successfully AI is being used.
- [14:30]
Interviewer / Host
I would say you mentioned, you know, Iranian state owned media outlets sort of being one of the actors in this. But is this like widespread and are there other bad actors that you, you know, have been able to verify?
- [14:43]
Disinformation Expert / Journalist
It depends on your definition of bad actors, I guess.
- [14:46]
Interviewer / Host
Yeah.
- [14:46]
Disinformation Expert / Journalist
The people, the accounts who have a blue check mark on X and who are spreading AI purely to get to the top of the feed and get interaction and therefore earn money by that. They are, and they are using us a lot. Iranian state actors are using it. They are sharing realistic, you know, missile attacks on Bahrain to claim that, you know, they are being more successful than they, they claim to be. They're showing US troops being captured by Iranian forces.
- [15:15]
Akilah Hughes
Yeah.
- [15:16]
Disinformation Expert / Journalist
But then they're also showing not very convincing AI of a group of Iranian soldiers sitting in a cave putting together weapons. Which just looks ridiculous, right? Well, it looks kind of photorealistic. Just the, the context of it seems ridiculous and I think that's clearly something people would see as AI. So they're still not getting it perfectly right and how they use it, but it's a lot better. Like all, everyone, I would assume every state, like we talked about, the US Administration Using AI. We, we've, over the years, Russia has been one of the most kind of technologically advanced countries that has tried to use disinformation to push its own narrative and disrupt politics in other countries. So they are also using AI. Not maybe in the same way as we've seen with the Iranians, but they are using AI and they're constantly developing their techniques. So I think in the coming years, as new events like this break out across the world, we're just going to see more and more and more of us.
- [16:16]
Interviewer / Host
Unfortunately, there was reporting today in the
- [16:19]
Akilah Hughes
US that, you know, we're not getting
- [16:20]
Interviewer / Host
clear numbers about those injured in or Iran from the military, which is a departure from military tradition in the U.S.
- [16:29]
Akilah Hughes
how do you think that sort of,
- [16:31]
Interviewer / Host
you know, I guess like factual gulf where it's like we don't know what's actually happening.
- [16:36]
Akilah Hughes
How do you think that factors into
- [16:37]
Interviewer / Host
people sort of coming up with their own answers via AI?
- [16:40]
Akilah Hughes
Like, do you think that that is
- [16:41]
Interviewer / Host
going to encourage more people to just make up what they believe is happening? Or, you know, does that impact the usage at all?
- [16:48]
Disinformation Expert / Journalist
No, I think it absolutely impacts. It's like if you have a void of information, people are going to fill us and they are going to look for information somewhere. And if the Pentagon is not providing accurate information to traditional news outlets, not to mind, you know, that the fact that it shut out traditional news outlets from its press corps, if it's not providing that information, then people will turn to Instagram or TikTok or X or Facebook where they might try to see if there is information out there. And especially if they are, you know, in one or other echo chamber online, if they are following certain influencers, those influencers are going to push the narrative that they have been pushing anyway, whether it's about, you know, the fact the US shouldn't be in the war or the US should be in the war. And so people are going to look to places for information that they shouldn't really be looking because they can't find it anywhere else.
- [17:41]
Akilah Hughes
Yeah.
- [17:41]
Disinformation Expert / Journalist
And so as a result, people are just going to be misinformed on both sides of the divide. I think it's not just one or the other. I think if you just don't have information and you want it, you will go and look for it somewhere. And you know, the, the Internet is not the most reliable place to find information, especially on social media platforms.
- [17:59]
Akilah Hughes
And the question becomes beyond the fact that they should, do these platforms really have an incentive to stop the spread of this?
- [18:06]
Interviewer / Host
Has anybody like made the move to be like we are going to morally be correct here and label things out of a duty to truth or anything of that nature.
- [18:16]
Disinformation Expert / Journalist
No, I assume it's look, for a year now or since the inauguration, every social media platform has just gone in reverse. They've backed out of everything that they, the progress that they made in terms of moderation and keeping their platform safe and you know, disinformation off platforms. That's gone now. Yeah, there are like X made that kind of vague, you know, wave. The San said, oh, we'll demonetize these accounts. I asked them how many accounts is demonetized. It won't tell me. Meta's oversight board, which is a kind of semi official board and has a bit of power but doesn't really have power, criticized Meta for its allowing AI disinformation on its platform during times of crisis. Meta said it welcomed the ruling and will maybe do something, but ultimately they have no reason to do something because they are not being held account by anyone. The US government was the only one who potentially had the power to do it and they are absolutely not going to do it because all of these companies have, you know, bent the knee and pledged allegiance to the Trump administration. It's happening in Europe. Europe is trying to hold them to account, but the most really they can do is to find them. And these companies have deep pockets, so that's not really going to make them change the way they do business.
- [19:32]
Interviewer / Host
Right.
- [19:33]
Disinformation Expert / Journalist
So unfortunately, I think the answer is no. No company has put their hand up and said we're going to do better and no one is there to make them do it.
- [19:41]
Akilah Hughes
The Meta oversight board that David mentioned said that the platform was, quote, neither robust nor comprehensive enough to handle the system scale and speed of AI generated misinformation, particularly during crises and conflicts.
- [19:55]
Interviewer / Host
So I mean, with that in mind, like all of these, all of these platforms have basically thrown up their hands and been like, well if somebody stops me, I'll stop. But till then, you know, I guess in your opinion, are we just cooked? Like, is it over? Is it just like every, we won't even know if our neighbors are AI. Like what is the future here?
- [20:15]
Disinformation Expert / Journalist
I look, I don't know, I, I, I think about it a lot and I don't know, like, I think people ultimately want authentic experiences online. Like some people, yes, one are happy with AI chat bots and will talk to them and whatever, but I think for most people they want authentic interactions. And as good as AI is at certain things, it's not Authentic. So I think ultimately we may see pushback against this. If X or Facebook running get overtaken by AI, people will just leave and look somewhere else. I think what we might end up seeing, and we're seeing a little bit of it already, is this very. Is this move away from centralized platforms where everyone is on them and you have to be on Facebook, you have to be on Instagram, and you will move into little communities online where people feel connected to people that they can connect to, you know, like Reddit, subreddits, that's that kind of thing, where people will no longer feel that they have to be on a mainstream platform and just want to be with the people that they feel most comfortable with. But that's going to take time and you know, it's gonna. It's hard to say if AI is gonna affect that as well. I don't think we're cooked quite yet. Getting there. We're. We're kind of like the frog in the water and it's getting to boiling point, but it's not quite a boiling point yet. We still can jump out. Yeah, I'm hoping we as a species will, but I don't know.
- [21:40]
Akilah Hughes
If spotting what's fake seems like an uphill battle, consider the challenge of understanding what is happening on the ground. Will this change the way that we witness war?
- [21:51]
Interviewer / Host
How do you think this will change the future of war coverage? Do you think that we can just expect. Maybe it's all coverage of everything, but
- [21:57]
Akilah Hughes
do you think that we can expect
- [21:59]
Interviewer / Host
more of this sort of fabricated, fake, you know, chest beating from all governments in the future? Or will we jump out of the water?
- [22:08]
Disinformation Expert / Journalist
I think, like to a greater or lesser degree, yes. I think we will see citizen journalists who have been empowered by the state that they're in to produce content that looks like it's on the ground, reporting on, when in fact it's not. It's AI.
- [22:22]
Interviewer / Host
Yeah.
- [22:22]
Disinformation Expert / Journalist
And it will get to the point where it'll be impossible to tell. And that that's the real worry, I think, where like you, you post a video like that of something happening, something egregious happening on the ground that no one else is covering, and you post it on social media and it goes around the world instantly and people get angry instantly and people respond instantly. And like a day later someone will have proved that it was AI, but it will be too late.
- [22:51]
Interviewer / Host
Too late. Yeah.
- [22:53]
Disinformation Expert / Journalist
And that's the real danger, is that there's no nuance, there's no patience anymore. People don't take the time it's instant. There has to be reaction. You have to respond. And that's. That's incredibly dangerous.
- [23:07]
Akilah Hughes
The AI flood is changing social media in the ways we see ourselves. When videos of battles, missile strikes, or troop movements are being generated and spread by AI before anyone can verify them, it erodes trust. It erodes truth. And that loss isn't just gonna be a casualty of war. We are already living in a time where the burden of proving truth is more pressing than ever. But with real lives on the line, we can expect that there will be grave consequences for this kind of behavior. And it's not really up to us. These platforms need to be regulated, need to care about their impact on the world. And that means people with real power coming together to force their hand. So too long didn't read. It's another instance where things aren't better. How Is this Better? Is a production of Courier. It's written and hosted by me, Akilah Hughes. It is produced by Devin Maroney. Video editing is by Shane Verkist. The rest of the team at Courier includes Kevin Dreyfuss, Sam Hollows, Marianne Kuga, and Charlotte Robertson. Please subscribe to follow how is this Better? On all the platforms YouTube, Apple, podcasts, Spotify, etc. And tell someone about your favorite episodes. If you're interested in sponsoring episodes or giving us products to try and try to sell, reach out to advertiseurier news room.com thanks for listening and until next time. See ya.
- [24:45]
Evan Osnos
Right now, news and politics are moving awfully fast. It can feel overwhelming, to say the least. I'm Evan Osnos, a staff writer for the New York on the Political Scene podcast. We slow things down to understand how power really operates in Washington, D.C. and what it means for you. My co hosts Jane Mayer and Susan Glasser and I have decades of reporting experience, and every Friday we have conversations with insiders and experts to understand the forces remaking America. Join us Fridays for the Washington Roundtable from the Political Scene. On Mondays and Wednesdays, you can also hear insightful episodes from our New Yorker colleagues David Remnick and Tyler Foggit. Available wherever you get your podcasts.