Return to Transcripts main page

The Situation Room

Anthropic: Claude Chatbot Working to Build its Own Successor; Inside the Online "Burnerverse" That Exploits Young Women. Aired 11:30a-12p ET

Aired September 18, 2026 - 11:30   ET

THIS IS A RUSH TRANSCRIPT. THIS COPY MAY NOT BE IN ITS FINAL FORM AND MAY BE UPDATED.

[11:30:00]

WOLF BLITZER, CNN ANCHOR: Happening now, another D.C. institution is now fighting to stop President Trump from permanently planting his name on the front of their building. Ousted board members of the U.S. Institute of Peace filed an emergency motion to stop the planned engraving. The latest battle comes after Trump dismantled the independent nonpartisan agency last December and renamed it after himself. We have reached out to the White House for comment.

And prosecutors and defense attorneys gave their opening statements this morning in the murder trial of former American Idol contestant Caleb Flynn. He's accused of killing his wife, Ashley, at their home just north of Dayton, Ohio back in February. Prosecutors say that Flynn, who is a former pastor, shot her, then tried to stage the scene to make it look like an intruder did it. And yesterday, prosecutors asked if jurors could set aside personal feelings about infidelity and whether they could consider a trial without a murder weapon. Flynn has pleaded not guilty. The trial is expected to last about two weeks.

And terrifying moments in the sky over Las Vegas. A passenger jet was forced to make an emergency landing when the engine started shooting flames just minutes after takeoff.

(BEGIN VIDEO CLIP)

UNIDENTIFIED MALE: Volaris 1741 appears a right engine fire. Do you have any issue?

UNIDENTIFIED MALE: Right. 1741. Mayday, mayday, mayday. Engine two failure. Request send back to the airport.

(END VIDEO CLIP)

BLITZER: The Volaris flight was heading to Mexico at the time of the incident. You can see bright orange flames shooting repeatedly from the plane's right engine. Fortunately, the pilots were able to circle back and land the plane safely. There's no word yet about what caused the incident, but the FAA is investigating.

PAMELA BROWN, CNN ANCHOR: Well, Wolf, new this morning, Anthropic says its A.I. agent, Claude, is helping to build the next version of itself. The artificial intelligence company said Claude currently leads more than a quarter of its research and development. That's up from zero from just months ago back in February. So, why does this matter?

Well, according to Anthropic, quote, "Models accelerating their own development could make it more challenging for humans to understand or control these systems."

BLITZER: And this comes just days after Anthropic's CEO called for an industry-wide showdown of the technology's development, writing, and I'm quoting now, "Since roughly this summer, A.I. has been advancing drastically faster, driven primarily by A.I.'s growing ability to build the next generation of A.I."

And with us now is Devin Kim, President of the Center for A.I. Safety. Good morning, Devin. Thanks very much for joining us. I want to start with this news from Anthropic. As I just mentioned only last week, the company's CEO warned about this reality. Put this into perspective for us. What are the risks when A.I. is helping build itself?

DEVIN KIM, PRESIDENT, CENTER FOR A.I. SAFETY: Thanks, Wolf. What people need to understand is that there's a very large gap in the public's understanding and the actual reality of this technology. The situation that we are-- we're in now is two years ago, 30 to 40 percent of a lot of the usage of the A.I. company's models were simply for helping students with homework.

[11:35:00]

Fast forward to today, you're seeing news about an A.I. solving century-old problems in mathematics, and at the same time, you're hearing news about these A.I. models breaking out of their sandboxes, the A.I. companies losing control of their own models, and having a lot of unintended consequences of that, including hacking into other companies.

And so, these are very serious problems, and what we're talking about here today is A.I. safety. How do we keep this very powerful technology safe for everyone so that we're able to make use of the opportunity and its benefits? And in doing so, how do we also avoid a concentration of power, because who will ultimately be in control of this technology? We're not wanting any small group of individuals or companies or governments to be making the decisions for all of us. That's very powerful and very consequential technology.

BROWN: And you say that these are very serious problems. What are the tangible impacts right now in this moment when it comes to A.I. and its impact on society, and what is your concern about what could happen in the future if it's not reined in?

KIM: So, what the Center for A.I. Safety is seeing in our research and also starting to see a lot of companies repeat for themselves as well is that the current capabilities of the models is to the level of which of expert-level hackers. This is part of OpenAI's recent release in the Astra model card, and you're seeing basically every single company admitting that they are losing control of their models and that they have all independently discovered that their models are actually hacking into critical infrastructure and other companies. And the consequences of this could be very severe. One can imagine hostile governments, terrorist organizations, or criminal groups, if they were also able to gain access to this technology, then they will be able to orchestrate large-scale attacks on critical infrastructure within the United States at a scale which has never been seen before. And these are very, very concerning things.

And the fact of the matter is that the current state of the United States, whether it's hardening our security system such that the risks from this can be mitigated, or the regulations that we have in place to actually align the incentives of these A.I. companies to be developing this technology responsibly, we are very, very far behind from the point at which we need to be today.

BROWN: So, I want to ask you about something that Anthropic CEO Dario Amadei, who is also one of the A.I. leaders sounding the alarm now, saying that there needs to be more regulation. This is what he told CNN's Anderson Cooper. Let's listen.

(BEGIN VIDEO CLIP)

DARIO AMODEI, CEO, ANTHROPIC: So, you know, the nightmare scenario for six to 12 months would be that you have, you know, a swarm of these agents, you know, that form some kind of botnet collective and, you know, could take down large segments of the internet. And then as the capability increases, it gets worse from there.

(END VIDEO CLIP)

BROWN: How realistic is that scenario? And how do we approach cybersecurity in light of A.I.'s current capabilities? I mean, is it too late?

KIM: It's very realistic. We are seeing evidence of this happening today. We are seeing these A.I. models breaking out of containment and harming things that we care about in our country and in the world today. There are actually quite a lot of things that can be done about this. The key principle to remember is we want to acknowledge how powerful this technology is. And the way forward is to think about how do we reduce concentration of powers in individuals, companies, and governments such that we can actually make sane decisions about this technology.

A thing that the Center for A.I. Safety has been advocating for a long time is liability law. We need to actually keep these A.I. companies accountable for the catastrophic risks and harms that they're imposing onto everyone. And our current set of liability laws are actually deeply inadequate. And there are actually two reasons that this is the case.

The first reason is that because of the rapid pace of this technology. And the second reason is that the current laws were not written with autonomous agents, autonomous A.I.s are able to independently act in the world in mind. And this has a lot of very severe consequences.

For instance, a lot of the issues of liability and holding people accountable really comes down to negligence. And negligence is actually very, very difficult to prove given the fast nature of this technology, given how unpredictable these things are, and given that we've already proven that we're actually not able to keep these models under control. And so, there has to be a lot of work done here.

A thing that people also often forget is that these A.I.s, these A.I. models, they actually run on very specialized hardware. They run on A.I. chips. The United States has the capacity to monitor and verify where these chips are and what exactly is that they're doing.

[11:40:00]

And we are not doing this, and this is something that we're actually able to do. And so, there's a lot of work ahead of us. We need to robustify security in society. We need to make sure our systems are safe. And we need to sort of move very quickly in order to make sure that we're actually doing um, have sort of like the sensible things in place such that we're actually able to, to direct appropriately.

BLITZER: Devin Kim. Thanks very much for your expertise and thanks very much for joining us.

BROWN: All right. Thank you. Coming up, a Situation Room special report on a deeply disturbing online network of anonymous men targeting and harassing college age women. I'm going to speak to the reporter who uncovered this network about how it operates and why it's so difficult to shut down. That's next.

(COMMERCIAL BREAK)

[11:45:00]

BROWN: Now, to a Situation Room special report. There is an online ecosystem built up of anonymous accounts on social media platforms like X and it's called burnerverse. Aside from talking college sports, the accounts target young women, exploiting them by sharing personal information or explicit images without their consent.

Joining us now a senior culture writer at wired EJ Dickson. She has been months investigating the so-called burnerverse. EJ, thanks for coming on really important work that you did here. Tell us, first off, what exactly is the burnerverse and how did it evolve from sports talk to something more malicious?

EJ DICKSON, SENIOR CULTURE WRITER, WIRED: Thank you so much for having me. Yes. The burnerverse is an online community of anonymous accounts and ostensibly their college sports fan accounts. They talk about the coaches, the players, the new recruits, but there is a corner of this community and the subculture that has emerged that is pretty aggressively racist and misogynistic and antisemitic.

And one of the things that they've started doing in recent years is doxing young college aged women and amplifying rumors or nonconsensual intimate images of them.

BROWN: And you actually spoke with some of the young women who have been targeted by these accounts. What stood out to you about their stories and what kind of impact has this experience had on them?

DICKSON: It's an immeasurable impact. I spoke to a young woman named Maddie Kowalski who had this happen to her last November. She had to withdraw from school. Her hair fell out. She developed rashes on her face. She developed PTSD. People were placing harassing phone calls to her parents that all hours of the night, it completely consumed her life. And she was not able to break away from it at all. It just followed her wherever she went.

BROWN: So, sad. I reached out to Maddie. She declined an interview understandably, but this is clearly caused so much trauma in her life and all of these other women who have been impacted by this so-called burnerverse. How big is this network of anonymous accounts?

DICKSON: It's difficult to quantify because it's very wide ranging and amorphous. It's kind of more accurate. Each college that has a big sports team kind of has its own individual online community. So, each burnerverse kind of has its own individual burner galaxy as it were. So, it's difficult to quantify and the accounts tend to be pretty small. They tend to be at most a few thousand followers.

But one of the things that was really shocking to me in my reporting is that I'd assumed that these guys were mostly frat kids or currently enrolled college students, but it's often men who are 10 or 20 or 30 years out of college with wives and kids from all over the country. It's really not limited to one specific demographic. And that's kind of what makes this so terrifying.

BROWN: It's really gross. Get a life. I'm sorry. I'm just like, hello, you're out of college. You have a family. Why are you sharing these images of young women and ruining their lives? What about accountability here? Are there laws on the books to help crack down on this? Why has this ecosystem been so difficult to stop?

DICKSON: There are, but it's very, very difficult. Unfortunately, in my reporting, I found that the wheels of the criminal justice system are very slow for victims of cyber harassment. Sadly, it's very difficult for laws to keep a pace with the speed with which technology has developed. And in a lot of these cases, in Maddie's case, her images went viral nationally within just a few hours.

The federal TAKE IT DOWN Act does require platforms to remove non- consensual intimate images and victims can submit a takedown request to various social media platforms. But more often than not, it's like a game of whack-a-mole. One image goes down, 10 more will come up in its place.

BROWN: What do you hope people will take away from your investigation and these women's stories?

DICKSON: Well, I hope to see some real change done within the platforms, particularly X, which is primarily where the burnerverse organizes. X's algorithm is such that it amplifies content if it's getting a lot of engagement and it's famously pretty poorly regulated. So, there's a lot of leeway for these men to post racist or sexist or antisemitic content and to dox these women. So, I mean, in a perfect world, I would like to see more regulation.

But more than that, I just want these men to know that the women that they do this to are brave and strong and will pursue accountability. You know, as Maddie's story demonstrates to all ends of the earth and anonymity is not a cover for doing and saying terrible things to somebody.

[11:50:00]

BROWN: Yes. Let it be known. EJ Dickson, thank you so much. We'll be right back.

(COMMERCIAL BREAK)

[11:55:00]

BLITZER: In the new CNN original series, "The Detour," the acclaimed chef, David Chang, steps away from online recommendations and algorithms and sets out with his friends in search of his next great meal.

BROWN: And along the way, he follows local leads, trusts his instincts, and of course, takes a few detours. Here's a preview.

(BEGIN VIDEO CLIP)

DAVID CHANG, CHEF AND HOST, "THE DETOUR": Let's define a road trip. How long does it have to be? I'm going to suggest the road trip has to be longer than a really long movie. It's got to be longer than my "Godfather II."

UNIDENTIFIED MALE: I agree with that.

CHANG: Anything less is a day excursion.

UNIDENTIFIED MALE: Yes. How much caffeine does this have in it versus a Red Bull?

CHANG: Don't worry about it. Let your heart guide you. Get in a flow state. Trust your surroundings.

UNIDENTIFIED MALE: I'm just taking whatever is given to me. I think added to road trip rules is like saying yes. Just trying new things, like just going down unexplored paths, following some poor advice from some stranger that you meet.

CHANG: And the reason I go to Las Vegas, it just is a whole different world that follows its own rules. And there's a little-known fact. They're handing out free money there.

UNIDENTIFIED MALE: You just got to be willing. You got to be brave enough to take it.

CHANG: Exactly. Sometimes that free money only happens when you lose some money.

(END VIDEO CLIP)

BROWN: Well, chef and host of "The Detour," David Chang, joins us now. So, tell us more about this show, David. You put down your phone, you got in a car and you go looking for great food without the internet's help, basically.

CHANG: Absolutely. Crazy pitch. And I can't believe the show actually happened because I wanted to show the world and remind people what it was like to find delicious things before the Internet, before smartphones. And it wasn't that long ago. And we sort of just asked people where to go and just got lost and threw away all the work in preparing the show and had a great time. Yes.

BLITZER: Yes, it looked like you had a great time. Why did you decide to do this as a road trip specifically?

CHANG: Because a road trip is something that is experiential. And I think it's one of the few things that everybody sort of can experience. They've done it themselves and it's just a lot of fun and there's a lot of great food along the way and great laughs. And that's what the show sort of encapsulates as well.

BROWN: Was there any detour that stood out in particular, something memorable or where you just couldn't believe the food?

CHANG: I would say all the episodes are fantastic. But the greatest detour we took was with comedian Eric Andre, who we weren't even planning to go to Miami. And somehow, we wound up in Miami and our producers and everybody involved with the show are so excited that we decided to legitimately make a detour to a city we didn't plan on going.

BLITZER: So, just give us a clue. Is there a meal, a food you discovered out there talking to these folks on this road trip that you're going to share with all of our viewers?

CHANG: 100 percent. One of the greatest culinary moments of my life was in Las Vegas. We were there with Court Jefferson and we were in a cab and the cab driver said, you should check out Ethiopian food. And I didn't even know that there was Ethiopian food in Las Vegas. And it's a longer story. It's an amazing story. You should watch the episode. But we had one of the best meals of the entire season at a beautiful Ethiopian restaurant.

BROWN: Did you invite the cab driver with you?

CHANG: That's a story in and of itself.

BROWN: OK.

CHANG: She was almost like a guardian angel that came to make sure that we were going to eat something delicious.

BLITZER: Give us a clue about the Ethiopian food. CHANG: She left a message with the restaurant saying that we would most definitely visit them the next day. And somehow, we did. And we had great, great food. The injera bread was fresh and delicious. And if you haven't had Ethiopian food, that's another example of the show is just to just get lost in a good way and try things that you may not have had or you always wanted to try. And this is the opportunity to do it.

BLITZER: Looking forward to it.

BROWN: It's important to when you choose your road trip partner, especially on these long drives. How did you choose who to drive with?

CHANG: You know, it's an easy ask when I can call a friend and say, hey, this is going to be all expense paid trip. You just have to spend time with me. But we're going to be able to eat anything you want, whatever you want to do, you can do. And thank you, CNN.

BLITZER: Well, we're looking forward to this program, David Chang. Thank you very, very much for doing it. It's going to be fabulous. Be sure to tune in a new episode, a new episode of the CNN original series, "The Detour with David Chang." It airs this Sunday, 9 p.m. Eastern.

And to our viewers and our listeners, thanks very much for joining this morning. You can always keep up with us on social media, @wolfblitzer and @pamelabrowncnn.

BROWN: And we will see you back here next week, 10:00 Eastern for two hours. Inside Politics with our friend and colleague Dana Bash starts right now.

[12:00:00]