The GoLaxy Papers: Inside China’s AI persona army

Summary

Podcast Summary: Click Here – The GoLaxy Papers: Inside China’s AI Persona Army

Host: Dina Temple-Raston
Guests: Brett Goldstein & Brett Benson, Vanderbilt University
Date: September 19, 2025

---

Overview:

This episode dives into the discovery of a trove of leaked documents—dubbed “the GoLaxy Papers”—revealing a highly advanced Chinese campaign using generative AI to create convincing digital Personas for targeted propaganda and cognitive warfare, including operations against the U.S., Taiwan, and Hong Kong. Host Dina Temple-Raston interviews two Vanderbilt University researchers, Brett Goldstein and Brett Benson, about how they uncovered the leak, what it reveals about the methods and ambitions of China’s state-backed influence operations, and why these developments mark a new frontier for information security and national defense.

---

Key Discussion Points and Insights

1. Discovery of the GoLaxy Papers

• Initial Tip-Off:
  • Brett Goldstein was alerted by a trusted researcher to check out a suspicious link containing a cache of Mandarin-language PDFs. Despite initial reluctance, he followed up, scanned for malware, and accessed the documents.

  “I go in and, hello, There's a whole bunch of PDFs on a public URL. And it's intriguing.” — Brett Goldstein (01:21)

• First Impressions:
  • The documents included technical schematics and photos of prominent Americans, signaling this was not spam, but a significant leak.

  “When you look at the docs, you realize that they had been working on this for years.” — Brett Goldstein (01:57)

• Translation Challenge:
  • Goldstein had to painstakingly translate the documents using an LLM model (“Llama”) over hotel WiFi, slowly uncovering their significance.

2. Nature of the Operation

• GoLaxy: China’s Persona Engine
  • The operation is run by a company called GoLaxy, connected to the Chinese Academy of Sciences and state security, with nearly 1,000 employees.

  “They claim in the documents that they serve Chinese national security and national strategy, interests and goals...and have a connection with the Chinese government.” — Brett Benson (09:21)

• Manual to Weaponize AI:
  • The leak contains a playbook for building psychological models on political targets, executed previously in Hong Kong, Taiwan, and potentially in the U.S.
• Scale of Targeting:
  • GoLaxy allegedly compiled dossiers on 2,000 U.S. public figures, thousands of right-wing influencers, and at least 117 Republican members of Congress.

  “It was pretty disturbing...profiles on thousands and thousands of targeted figures in the United States...screenshots...proof of concept.” — Brett Benson (10:34)

3. The Technology and Its Evolution

• Beyond Basic Bots:
  • Unlike older online propaganda, these Personas are not generic bots but fully realized digital identities, tailored by scraping LinkedIn, Twitter, Instagram, etc.
  • Personalization:

    “What we're seeing with artificial intelligence and generative AI is you can really create detailed Personas tailored to the individual.” — Brett Goldstein (06:44)

  • Made to Deceive:
    • Personas not only parrot talking points but listen, adapt, argue, and comfort, making them much harder to identify as fake.
• Data Harvesting at Scale:
  • GoLaxy scrapes millions of data points from platforms—Facebook, X (formerly Twitter), Weibo, WeChat, Reddit—to build behavioral and psychological profiles.

  “They build these profiles based on user interactions and open source platforms.” — Brett Benson (15:24)

• AI-Driven Content & Engagement:
  • China’s ChatGPT competitor, DeepSeek, powers the system. AI Personas can engage authentically, influence discussions, alter the direction of conversations, and subtly seed or amplify narratives.

  “They're really subtle, realistic, human like in their ability to engage and then to see content and to shape narratives.” — Brett Benson (16:50)

4. Comparison with Global Influence Operations

• More Subtle Than Russian Troll Farms:
  • Unlike Russia's approach in 2016, which used blunt-force bot networks, China's Personas are intricately designed for individual influence.
• “TikTok on Steroids”:
  • GoLaxy's system personalizes at a deeper level—whereas TikTok targets based on general interests, these Personas are engineered to interact with specific targets and shift political opinion.

  “What goalaxy is doing...it's using politically aligned AI ops to target users with...strategically framed content.” — Brett Benson (17:11)

• Speed and Scale:
  • The coordination and deployment of influence campaigns can occur at a pace that outstrips anything humans could accomplish.

  “The data harvesting and the content creation, the scale and speed can far eclipse Anything that TikTok can do by several orders of magnitude.” — Brett Benson (18:08)

5. The Data Question: “The Rainy Day Has Come”

• Historical Data Breaches:
  • The podcast references past breaches—OPM, Anthem, Starwood—and suggests we may now be seeing their purpose realized: fueling AI-driven cognitive warfare.

  “This was the kind of aha moment for me when I realized that the data and the data mining efforts paired with AI has the capacity to engage in sort of a front of cognitive warfare that I didn't really think about before.” — Brett Benson (18:54)

• A New National Security Frontier:
  • AI-driven influence ops represent a new type of coercion, with implications for national defense, deterrence, and democratic legitimacy.

6. Urgency and Public Disclosure

• Why Go Public:

  “My philosophy from day one, after seeing that, is we need to get this out the door as soon as possible.” — Brett Goldstein (20:20)

  • Goldstein and Benson published the trove online unredacted to prompt defenses and academic scrutiny without delay.
• AI Changes the Game:
  • Building influence operation infrastructure that once took months can now happen in moments.

  “Now what do I just hit? Upload or ingest? And it's a game changer.” — Brett Goldstein (20:44)

7. Detection, Defense, and the Disinformation Crisis

• Undetectable Fakes:
  • Current defenses can spot botnets; these individually crafted Personas are almost undetectable. If platforms can’t distinguish AI from human, what happens to truth?

  “This is a new animal, and I'm not sure it's detectable.” — Brett Goldstein (21:21)

• Strategic Implications:
  • The potential undermining of deterrence and the integrity of military and civilian decision-making.

  “It may be the case that these information operations undo some of the benefits of AI.” — Brett Benson (22:10)

• GoLaxy Attempts to Cover Tracks:
  • When exposed by the New York Times, GoLaxy scrubbed evidence from its websites in near real-time, despite issuing denials.

  “Pages were disappearing...Like there was a whole section on government relationships and that just disappeared.” — Brett Goldstein (23:24)

8. The Takeaway: “It’s Here”

• Immediate Relevance:
  • The researchers stress the urgency: AI-generated propaganda and manipulation have already arrived—they are not a concern for the distant future.

  “AI generated propaganda, AI generated information manipulation is here...That is here. That's not a science fiction future. That is here.” — Brett Benson (24:33)

---

Memorable Quotes & Timestamps

• “It turns out it's talking about a sentiment system in China. So someone is trying to convince someone else to have a certain view.” — Brett Goldstein (04:45)
• “Not just bots mindlessly retweeting slogans, but carefully constructed identities. Digital doubles.” — Dina Temple-Raston (05:22)
• “The difference is that TikTok is targeting users with lifestyle and entertainment content... What goalaxy is doing...it's using politically aligned AI ops to target users with...strategically framed content.” — Brett Benson (17:17)
• “If we can't solve this and we don't know human versus machine, we're in a lot of trouble.” — Brett Goldstein (21:40 & 24:03)

---

Key Timestamps

• [00:26 – 01:32] Discovery of leaked documents
• [04:45] AI sentiment system revealed in the papers
• [06:44] New era: hand-crafted, AI-powered Personas
• [09:06 – 09:55] GoLaxy: company background and connections
• [10:34] U.S. figures profiled by GoLaxy
• [15:02 – 16:50] Deep-dive into data harvesting and AI Persona creation
• [17:07 – 18:08] TikTok comparison and unique threat model
• [18:54 – 20:08] The “rainy day” of data exploitation arrives
• [20:20 – 20:44] Rationale for quick public disclosure
• [21:16 – 21:56] Detection challenges and existential stakes
• [22:10 – 23:11] Implications for military strategy and AI trust
• [23:24] GoLaxy’s attempted cover-up
• [24:28 – 24:33] Urgency: “It’s here.”

---

Final Takeaways

• Generative AI is already powering highly personalized, nearly undetectable online information operations targeting individuals at scale.
• China’s state-backed operations may be the “new frontier” of cognitive warfare, weaponizing massive data troves and AI for strategic influence.
• The defensive playbook is not yet written. Detection and attribution challenges threaten societal trust and national security.
• This isn’t science fiction—the threat is now.

---

For listeners and policymakers: The GoLaxy Papers shine a spotlight on a paradigm shift in information warfare that will shape the future of democracy, security, and digital life.

Transcript

[Dina Temple-Raston]

From Recorded Future News and prx, this is Click Here's Mic Drop. A longer listen to one of our favorite interviews of the week. I'm Dena Temple Raston, and today, two Vanderbilt University researchers who got a tip in April that sent them down a rabbit hole they never imagined.

[Brett Goldstein]

My phone dings, and I look, and it's from a researcher who I know and respect, and he's like, you need to check this out.

[Dina Temple-Raston]

Brett Goldstein is a special advisor for the Institute of National Security at Vanderbilt University. And his first instinct when he got that message was to ignore it.

[Brett Goldstein]

And so I'm like, huh? So I then forget about it.

[Dina Temple-Raston]

In Brett's defense, he was at a conference in Australia at the time, and. And his hands were full. But when he finally got back to the hotel, he scrolled through his messages again. The same note. You need to check this out.

[Brett Goldstein]

It reads, and I look at it, and it's a link, and I hate links.

[Dina Temple-Raston]

So he scanned it for malware and ran some extra checks just to make sure it was clean. And then finally, he clicked.

[Brett Goldstein]

I go in and, hello, There's a whole bunch of PDFs on a public URL. And it's intriguing.

[Dina Temple-Raston]

What was intriguing?

[Brett Goldstein]

Well, it was all in Mandarin. And I start scrolling through pages, and what I'm seeing is Mandarin. I'm seeing technical schematics, and then I'm seeing pictures of prominent Americans.

[Dina Temple-Raston]

And that's when Brett realized this wasn't spam. This was a leak, and this was something big.

[Brett Goldstein]

When you look at the docs, you realize that they had been working on this for years.

[Dina Temple-Raston]

Stay with us. Support for Click here comes from CleanMyMac. If you're like me, your cloud drives are packed with files you don't actually need. Syncduplicates, old backups, random junk, just gobbling up storage. CleanMyMac takes care of that with a new cloud cleanup feature. It connects to your accounts on iCloud, OneDrive and Google Drive and scans to find large space wasters, both in cloud storage and on your device. And while my Mac health is apparently excellent, CleanMyMac found 8 gigabytes of junk on my computer and tons of duplicate downloads. And it removed a bunch of leftover applications I never use. Who knew? And this happens locally on my Mac, so my data stays safe. CleanMyMac has a whole suite of other features, too, including their Moonlight Anti Malware Engineering. With just a click, the Anti Malware tool scans and removes viruses, ensuring your Mac remains threat free. Not everything deserves eternal storage, so why pay for it? Get tidy today with CleanMyMac. Try it free for seven days and use the promo code. Click to save an additional 20% on your [e-mail] you. I'm Dina Templewest and this is Click Here's Mic Drop. Brett Goldstein didn't remember where he was exactly when he started pawing through that cache of mysterious documents.

[Brett Goldstein]

I was somewhere in another Marriott, because it's always a Marriott, right? On some crappy wi fi. I'm like, I'm like, I need to go to sleep. I need to go to sleep. I'm giving a talk in the morning. I need to go to sleep. And I'm like, I need to see what it is.

[Dina Temple-Raston]

So he downloaded a large language model, or LLM one called Llama, and it was driving him crazy because he appeared to be on the world's slowest hotel wi fi.

[Brett Goldstein]

I remember being particularly irritated because I was trying to download a llama and it was a gabillion gigs and I'm just like, fuck.

[Dina Temple-Raston]

Hour after hour, he fed chunks of the Mandarin text into the model. He found schematics, PowerPoints, briefs, and what emerged was a kind of playbook.

[Brett Goldstein]

And it turns out it's talking about a sentiment system in China. So someone is trying to convince someone else to have a certain view.

[Dina Temple-Raston]

In other words, internal documents about a propaganda operation. And it wasn't just your average run of the mill Chinese campaign. This wasn't crude. This was intimate, surgical.

[Brett Goldstein]

And I start seeing mentions of Hong Kong and Taiwan, and I'm scrolling through it and I'm like, I think they're talking about pretty Personas here.

[Dina Temple-Raston]

Personas. Not just bots mindlessly retweeting slogans, but carefully constructed identities. Digital doubles. They looked real, they sounded real. They argue, they console, they slide into your feed and make you forget you ever doubted them. Brett recognized this immediately because, well, he's been studying Personas and the way people respond to them for a while now.

[Brett Goldstein]

A decade ago, I would create clusters of people. You'll say, Midwest, white, whatever politics, this economic bracket, you'd have 10, 20 variables. That creates a cluster. That was considered good enough back then.

[Dina Temple-Raston]

The recipe was simple. Pick a demographic, bake a Persona, set it loose. White guys in Ohio, single moms in Michigan, retirees in Florida. But what the Chinese documents seem to be describing, it was next level. Not generic stereotypes, not cookie cutter avatars. These Personas were handcrafted for you, one person at a time, and behind them, generative AI, which meant they didn't just parrot talking points, they listened, they adapted they. They whispered back in a comforting voice.

[Brett Goldstein]

What we're seeing with artificial intelligence and generative AI is you can really create detailed Personas tailored to the individual.

[Dina Temple-Raston]

The documents laid it all out, how they were creating very believable social media Personas with AI and then tailoring those accounts to interact not with a crowd, but with someone very specific. They did it, the paper said, by scraping LinkedIn profiles and Twitter feeds and Instagram posts, and then used that data to build a Persona that could slip seamlessly into someone's world. And According to the 399 pages of Chinese documents in the leak, this wasn't theoretical. It had already happened in Hong Kong and in Taiwan and maybe even here.

[Brett Goldstein]

When I saw photos of Americans in the dock, I'm like, oh, boy, this is going to be an issue. This is a threat, and we need to get this out as fast as we can to help educate people and start thinking about defenses.

[Dina Temple-Raston]

So Brett called in reinforcements.

[Brett Goldstein]

As you'll see as the story unfolds, it required more than one Brett to make this happen.

[Brett Benson]

My name is Brett Benson. I'm a professor of political science and political economy at Vanderbilt University.

[Dina Temple-Raston]

Brett Benson is one of Brett Goldstein's colleagues at Vanderbilt. He's a political theorist, a specialist in the whys and wherefores of armed conflict, and crucially, he speaks Mandarin.

[Brett Benson]

I speak Mandarin Chinese, lived in Taiwan, and have done research both there and in China. We were having a conversation about security in East Asia, and he said, I want to share something with you. And he said, I don't know what it is. It could be something cool, he said, but I just don't know.

[Dina Temple-Raston]

At first, Benson wasn't sure.

[Brett Benson]

Initially, I was very skeptical. Told him I didn't really know what to make of it, needed to spend some more time with it.

[Dina Temple-Raston]

But as he read through the documents, his heart started to sink.

[Brett Benson]

I said, it appears that there's a company in China that is conducting AI information ops, and they're interested in cognitive warfare targeting the United States, Taiwan, and Hong Kong.

[Dina Temple-Raston]

The company's name, Go Laxi. On paper, it looked like a private company, but it was clear that it had what the Chinese call guanxi were connections.

[Brett Benson]

In the documents, they claim to work with the intelligence services, both in the military as well as the sort of civilian intelligence services. They claim in the documents that they serve Chinese national security and national strategy, interests and goals. And then I also noticed that they claim to have a connection with the Chinese government.

[Dina Temple-Raston]

And the documents made clear that Go Laxi wasn't Just another startup. It was founded by the Chinese Academy of Sciences, a state run research center. And the documents seemed to indicate that the company had nearly 1,000 employees.

[Brett Benson]

It wasn't just a company that was operating independently, at least.

[Dina Temple-Raston]

And at its core, Benson realized Go Laxi was building psychological models at scale on political targets. The documents claimed the operation had been deployed in Hong Kong when there were protests about a Chinese backed national security law. It was used in Taiwan during elections. And it also claimed to have compiled dossiers on 2,000 US public figures, thousands of right wing influencers, and at least 117 Republican members of Congress. When you saw the pictures of the Americans, what went through your head?

[Brett Benson]

Yeah, that was a little bit disturbing. It was pretty disturbing. The documents claim to have developed profiles on thousands and thousands of targeted figures in the United States as well as in Taiwan and Hong Kong. And then there's these screenshots that was sort of like proof of concept. Look, you know, look at the information that we have on people that I know and see in the news.

[Dina Temple-Raston]

And as far as Brett could tell, this wasn't bravado. Go Laxi was showing their work names, faces, profiles, all drawn from our digital lives and then fed back into machines. And the two Brets then had the same thought. If Golaxy could do that with lawmakers and influencers, what could they do with the rest of us? When we come back, the two Brets look under the Golaxy hood and tell us how they think it all works. Stay with us. Support for Click Here comes from Express VPN. Express VPN stops hackers from stealing your data by creating a secure encrypted tunnel between your device and the Internet. It would take a hacker with a supercomputer or over a billion years to get past ExpressVPN's encryption. ExpressVPN is rated number one by Top Tech reviews like CNET and the Verge. I take my Internet security seriously and I know that Express VPN does too. It's easy to use and works on all your devices. You just fire up the app and you're protected. Secure your online data today by visiting expressvpn.com clickhere that's E X P R E S to find out how you can get up to four extra months free. Expressvpn.com clickhere Support for Click Here comes from Claude AI. Claude is the AI for minds that don't stop at good enough. It's the collaborator that actually understands your entire workflow and thinks with you, not for you. Whether you're debugging code at midnight or strategizing for your next business move. Claude extends your thinking to tackle the problems that matter. Who needs a regular search when you can have a conversation with Claude about what you're looking for? And unlike some of the other AI helpers I've used, Claude doesn't retain information from previous conversations, so it's private. Ready to tackle bigger problems? Sign up for Claude today and get 50% off Claude Pro when you use my link. Claude AI Claude, click here. That's Claude AI. Clickhere right now for 50% off your first three months of Claude Pro, and that includes access to all the features mentioned in today's episode. Claude AI Clickhere.

[Recorded Future Announcer]

In cybersecurity, your greatest fear isn't the threats you see, it's the critical signals lost in the noise. Every day, security teams sort through millions of potential threats. That's why Recorded Future exists, to give you precision intelligence tuned to your needs. Our advanced AI detects patterns humans might miss, while our threat intelligence experts, veterans of military and intelligence services, provide crucial context. With Recorded Future, you gain the confidence to identify critical threats and the precision to act before they become attacks. Learn why 1900 customers, including 45 plus sovereign governments, trust us to detect threats faster and and achieved 350% plus ROI within a year.

[Dina Temple-Raston]

As the two brets poured over the documents, the picture got more worrisome. This wasn't just about a handful of politicians or influencers. It was about a method, a playbook on how one might weaponize AI against whole societies. And step one was deceptively simple. Harvest everything.

[Brett Benson]

Goxy scrapes millions of data points from places like Facebook, X, Weibo, WeChat, Reddit, other online forums. And then it holds data on tens of thousands of individuals and organizations.

[Dina Temple-Raston]

So they're just aggregating data, millions of pieces of us, until they have an occasion to put them to use.

[Brett Benson]

And then the next step is to start creating behavioral and political like these psychological profiles of individuals and organizations and networks. And then they build these profiles based on user interactions and open source platforms.

[Dina Temple-Raston]

Take for example a user who loves the music of Dave Brubeck and routinely chats online with other fans. Golaxy could spin up a fake Brubeck devotee and have them meet online.

[Brett Benson]

These are fake but highly realistic, like you know, AI generated bots. It's a made up online profile basically that looks like a real person, interacts like a real person. Much more subtle than the Russian operations that we saw in the 2016 election. They're very, very tailored to appeal to a certain individual. The Personas show up in regular online spaces and they engage in friendly and relatable exchanges. And then they start to see content discussions, amplifying certain viewpoints, altering certain viewpoints, hijacking, trending topics, jumping into hot conversations that take place to subtly slant them in different directions.

[Dina Temple-Raston]

And unlike Russia's troll farms, goalaxy didn't need to have thousands of people tapping away at keyboards. It had something faster, cheaper, and endlessly adaptable. China's version of chatgpt, something called deepseek.

[Brett Benson]

The thing that really blew me away is the adaptability of the Personas, and that's the part that's kind of spooky to me. They're really subtle, realistic, human like in their ability to engage and then to see content and to shape narratives.

[Dina Temple-Raston]

Brett compares the operation to TikTok on steroids.

[Brett Benson]

TikTok tracks what people watch, right? What they like, what they share, what they re watch to learn the interests of the users. And, and then it uses the data that it gathers based on human interaction to produce more content for the users. So over time, its algorithm fine tunes to the individual's feed so that it's like really, really personalized and keeps users watching longer. The difference is that TikTok is targeting users with lifestyle and entertainment content as well as, like, commercial ads. What goalaxy is doing, it's using politically aligned AI ops to target users with like, politically and strategically framed content.

[Dina Temple-Raston]

And the TikTok algorithm steers users toward real human creators, while Golaxy's algorithm steers users to their AI creations, all at a scale humans could never replicate on their own.

[Brett Benson]

This is all AI driven. And so the data harvesting and the content creation, the scale and speed can far eclipse Anything that TikTok can do by several orders of magnitude.

[Dina Temple-Raston]

China has been vacuuming up all kinds of information for years. They broke into the Office of Personnel Management and Anthem helped back in 2015. Three years later, they cracked into Starwood Hotels. And we've always wondered what the Chinese would do with all that data. And the way it was always explained to me and people on the intel side was, you know, they're saving it for a rainy day. They know someday they'll be able to use it. Has it. Does this indicate to you that the rainy day has happened?

[Brett Benson]

Yeah, that was the, that was the big revelation to me because like you, I've had the same reaction to this massive collection of data. Face recognition all over China, censorship, data collection on its own citizens and but Then the big question in my mind has always been, okay, what's it for? Is it just to try to increase security domestically? Maybe abroad? This was the kind of aha moment for me when I realized that the data and the data mining efforts paired with AI has the capacity to engage in sort of a front of cognitive warfare that I didn't really think about before. I think this is the new frontier in national security because there are these different domains of coercion, everything from military operations to economic coercion, cyber. And now we have this information ops. Propaganda has always been part of information ops, but this is, this is completely different because of the scale and speed and the reach.

[Dina Temple-Raston]

Which is why the two Bretts have been so open about what they found. In September, they put the documents unredacted online for other academics to see. Here again is Brett Goldstein.

[Brett Goldstein]

Academically, there's an inclination to sit on some of these things for a couple years and really just research the heck out of it. But my philosophy from day one, after seeing that, is we need to get this out the door as soon as possible.

[Dina Temple-Raston]

Because AI wasn't just making the work of Personas easy. It was making it fast. Really fast.

[Brett Goldstein]

Back in the day, we'd spend a month creating a relational data structure. Now what do I just hit? Upload or ingest? And it's a game changer.

[Dina Temple-Raston]

You don't have to spend months cleaning the data or reformatting it. Now with AI, you can just dump in oceans of data and the model does the rest.

[Brett Goldstein]

AI is evolving at a speed we've never seen before, and I've never seen before, and there's just so much. And the frontier models are just crushing it.

[Dina Temple-Raston]

And they are moving so fast that defenses to stop them don't even exist yet.

[Brett Goldstein]

Like, we can find your typical bots and all the garbage that's out there, fine. This is a new animal, and I'm not sure it's detectable.

[Dina Temple-Raston]

AI is pretty good at spotting a bot targeting thousands. But a Persona created just for you? Maybe not.

[Brett Goldstein]

So how do we solve this? If we can't solve this and we don't know, human versus machine, we're in a lot of trouble.

[Dina Temple-Raston]

And if platforms can't distinguish real accounts from fake ones, what happens to the truth itself?

[Brett Benson]

What does this do to our perception of truth and to news and things like that? If I were to talk to Congress, I would start talking about the bigger implication of for national security.

[Dina Temple-Raston]

Not just voters, not just public opinion, but something even bigger than that.

[Brett Benson]

One of the things that we are, Brett and I are now starting to think about is how does this affect military operations? How does this affect, how do these types of AI generated information operations affect deterrence? One of the things that we all hope that AI would do is improve decision making between adversaries and thereby increase deterrence and make the world more peaceful. Because after all, it just sharpens decision making, makes it faster. But in a world where information can be manipulated like this, both at the public level, in democracies, among voters, as well as, you know, targeted fake content toward decision makers, soldiers, people in the military, and then importantly, manipulating the information itself that LLM models are trained on, it may be the case that these information operations undo some of the benefits of AI.

[Dina Temple-Raston]

This summer, as the two Brett's were preparing an opinion piece for the New York Times, this crazy thing happened. The paper contacted Go Laxi for comment and the company suddenly started deleting things on their website.

[Brett Goldstein]

Pages were disappearing. We were watching in almost real time and just thing after thing was being deleted off their site. Like there was a whole section on government relationships and that just disappeared. And you know, the pages would now be dead ended, the links would just all be dead.

[Dina Temple-Raston]

Golaxy denied everything, but the disappearing pages from the website seemed to suggest otherwise. Nowadays, Brett Goldstein is busy trying to learn how to detect these customized Personas.

[Brett Goldstein]

I just want to understand what's an AI versus a human. And I think that is a fundamental thing we need to do. So how do we solve this? Because the alternative to this is if we can't solve this and we don't know human versus machine, we're in a lot of trouble.

[Dina Temple-Raston]

And Brett Benson is trying to sound the alarm. What's the one thing you wish that more people understood about this story?

[Brett Benson]

That it's here.

[Dina Temple-Raston]

When you say it's here, what is the it?

[Brett Benson]

AI generated propaganda, AI generated information manipulation is here. When I use Chat GPT, I increasingly trust it. I think a lot of people increasingly trust it. But there are other forums where AI is being used to manipulate information, to sow doubt, to generate disinformation and disorientation. And that I think I wish people, more people interested that that is here. That's not a science fiction future. That is here.

[Dina Temple-Raston]

And maybe that's the lesson. Every new technology that promised to connect us has also been used to divide us. The printing press, the telegraph, the radio, and now generative AI, a far more sophisticated tool. And each time we've learned to adapt. So the question is whether this time we can adapt fast enough. From Recorded Future News, this has been Click Here's Mic Drop. It was written and produced by Megan Dietrich, Sean Powers, Erica Gaeda, Zach Hirsch, Lucas Riley and me, Dina Temple. Rest it was edited by Karen Duffin. We'll be back on Tuesday with an all new episode of Click Here. Have a great weekend. Looking for more of the cybersecurity and intelligence coverage you get on Click Here? Then check out our sister publication, the Record from Recorded Future News. You'll get breaking cyber news from reporters in New York, Washington, London and Kyiv, among others. And you'll see for yourself why it attracts hundreds of thousands of page views every month. Just go to the Record Media.