Return to Transcripts main page

CNN News Central

Trump Calls Concerns Over A.I. A "Hoax": A.I. Fears Fuel Growing Political Debate Ahead Of Midterms; Dems And GOP Lawmakers Split Over Calls To Regulate A.I.; Watchdog: "Alligator Alcatraz" Put Detainees In Metal Cages; States Sue Over Rule Making It Harder For Some To Get Green Cards; Airman On Damaged Parachute: "Most Terrifying Thing I've Ever Seen"; Air Force Col. Shot Down In Iran Shares Harrowing Rescue Story. Aired 3-3:30p ET

Aired September 14, 2026 - 15:00   ET

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

BRIANNA KEILAR, CNN HOST: And well-put, Don. Thank you so much.

DON RIDDELL, CNN SPORTS ANCHOR & CORRESPONDENT: Yes. All right.

KEILAR: A new hour of CNN NEWS CENTRAL starts right now.

Threat or Hoax: As tech CEOs warn that A.I. poses a clear and present danger to humanity, President Trump dismisses their warnings, suggesting it's a sick conspiracy.

And a daring rescue behind enemy lines. A U.S. Air Force officer sharing for the first time details about the risky recovery operation after he was shot down over Iran.

And new reporting on how some immigrants held at Florida's Alligator Alcatraz were sometimes held in small metal cages, sometimes for up to two hours.

We are following these major developing stories and many more all coming in right here to CNN NEWS CENTRAL.

As a debate over A.I. regulations consumes Washington today, President Trump revives a familiar refrain, posting, quote, "A.I. taking over the world, destroying humanity and all other things bad is a Hoax, no different from RUSSIA, RUSSIA, RUSSIA, UKRAINE, UKRAINE, UKRAINE - IMPEACHMENT HOAX number one, IMPEACHMENT HOAX number two, and all the other HOAXES and SCAMS."

Before this post, Democrats called for Trump officials to give a classified briefing after the nation's leading A.I. executives, including Elon Musk, publicly agreed A.I. advancements need pacing. CNN Congressional Correspondent Lauren Fox is on the Hill for us.

Lauren, what are lawmakers saying?

LAUREN FOX, CNN CONGRESSIONAL CORRESPONDENT: Well, right now you have this stark divide between Republicans and Democrats on Capitol Hill over what they gauge as the threat radius for artificial intelligence. And you have people like Speaker Mike Johnson arguing that this isn't really a time for Congress to take up dismantle, that this is an evolving technology, that there are National Security concerns when it comes to trying to slow down A.I., in part because he argues is China and the U.S. enemies going to be doing the same across the globe.

He also said that he does think it is worth having conversations, and he said that he does hope to soon meet with A.I. executives to have some of those conversations. Meanwhile, you have Democratic leader in the House, Hakeem Jeffries, as well as Senate Minority Leader Chuck Schumer, calling for further action and responsibility from Congress.

You heard from Schumer this morning that he wants to have a classified briefing for members of Congress to understand what is happening.

Now, the prospects for any legislation right now, Brianna, are slim, and that is because of the calendar. You have just a week that House lawmakers are in session right now, and then Senate lawmakers are only here until the end of the month. That doesn't give them a lot of time to iron out what is obviously very complex legislation that would touch almost every single committee in Congress, right?

This isn't one bill that goes through one single committee. This potentially could have an impact across the board because A.I. affects financial institutions. A.I. impacts national security. So, you can see there that this is a really broad and complex issue that a lot of lawmakers have been trying to sound the alarm on for several months.

I mean, I just looked today. There have been more than a hundred A.I.- related bills introduced in Congress in the last two years, and yet not a single one has passed, and there's been really no strong momentum at this point, in part because this is evolving technology and in part because there is starting to be a little bit of a partisan divide over just how far Congress wants to go here. Brianna.

KEILAR: Yes, we are starting to see that for sure. Lauren, thank you so much.

And as lawmakers and A.I. leaders are grappling with how and whether to regulate artificial intelligence, we're learning the threats posed by the technology are far from hypothetical. Last Thursday, Anthropic released a report outlining real-life case studies where users tried to use Claude for malicious activity. Anthropic says it has disrupted activity across seven harm areas, cyber operations, influence operations, surveillance, scams and fraud, biological misuse, conventional weapons development, and illicit distillation.

They said, quote, "The threat actors covered in this report include suspected state-sponsored groups, financially motivated criminals, commercial spyware vendors, state propaganda institutions, and politically motivated individuals."

In one instance, Anthropic says it identified and disrupted an Iran nexus threat actor that used Claude to collect and analyze publicly accessible data to develop targeting recommendations against U.S. naval forces in the region. In another investigation, Anthropic found a People's Republic of China government-aligned operation that used Claude to track, profile, and recruit Uyghurs and Uyghur armed formations in Syria. CNN has reached out to Chinese authorities for comment.

[15:05:02]

And in another instance, a single Claude subscriber used Claude to build a system that circumvented Mali's legal restrictions that require a court order to build a surveillance platform that monitors roughly 25 million SIM cards on all three of the country's national mobile operators.

Anthropic explains that the cases they share here aren't typical misuse, but rather examples of the most notable and novel threat activity that we've identified to date. The report warns as models become increasingly capable, their risks will increase unless A.I. developers and society's defenders act to make them safer. We're joined now by Stephen Witt. He's the author of "The Thinking Machine," a history of the A.I. giant NVIDIA.

Thank you so much for being with us. We just talked about this threat report from Anthropic. And in a separate essay published two days later, Anthropic's chief, Dario Amodei, warned specifically about two phenomena, recursive self-improvement and agent swarms. Can you help us understand what those are?

STEPHEN WITT, AUTHOR, "THE THINKING MACHINE": Sure. So agent swarms are when we spin up a thousand A.I.s at once and set them loose on a problem with very little extra guidance. So, we might have them solve a -- a tricky mathematical puzzle or something like that. What happened in the most recent incident for an open A.I. is actually an agent swarm escaped. It escaped containment, it made it onto the internet, and it hacked someone else's computer. And they were collaborating together, colluding together, and attempting even to disguise their actions from human overseers.

The swarms are very powerful. I mean, they can advance mathematics, as I said, they can advance medicine. They're going to be a tool we see a lot in the years to come, but they can also have the potential to go out of control. And that gets to the other thing you mentioned, which is recursive self-improvement.

And that just means the A.I. is making itself smarter. It's using the research it's doing to advance its own capabilities. And when you have a thousand A.I.s at once collaborating to do that, it might lead to an out-of-control situation where the A.I. is more powerful than we can control or maybe even understand.

KEILAR: Yes, it's scary. And OpenAI's Sam Altman said there's two ways that A.I. progress could go very badly. First, we could lose control of the future to AI. And second, we could end up in a world with too much concentration of power. You're calling for a global pause on A.I. research and development. What do you see as the window of action to regulate A.I.?

WITT: Well, to be clear, I'm just amplifying the call that they made for a global pause on A.I. development. This is something that's coming from OpenAI and Anthropic themselves. They think we're moving too fast. They want to slow down. And so I'm really just amplifying the things that they are saying.

But I think what they're worried about is that we're hitting this frontier of recursive self-improvement. I think they're looking at the tools that they themselves have developed and are terrified of them. And they just don't know what to do. They're essentially begging the U.S. government to regulate them. And that's on the verge of a trillion-dollar IPO.

And it doesn't seem like, at least the Trump administration, is too interested in doing that. They're rejecting the call for more regulation.

KEILAR: Yes, so they -- they are and you hear the President saying it's a hoax. China is going to benefit from this. You hear Trump administration officials saying you're just worried about liability issues or you just want to reduce competition in the A.I. space. What do you say to those charges against these A.I. companies who are asking for regulation? Is there any truth to those being motives for wanting it?

WITT: No, that's just conspiratorial thinking. These things really did break out of their container and they really caused havoc on the internet in a way that was totally unexpected and -- and really terrifying to a lot of people. That really happened. You know, the reporting on that is very clear. How you want to respond to that is kind of a different question, but it isn't some kind of hoax. You know, that really happened.

Now, I do agree with Trump on this point and Trump's administration. If we don't get China on board, there's no point. If we regulate A.I. just in the U.S. and China doesn't follow, we've actually put ourselves at a disadvantage and China will then develop a world conquering A.I. and it'll be just as dangerous, only it'll be coming from there and not here. So, that point is true.

President Xi is visiting Trump in the White House later this month and this gives Trump really a -- a unique and historical opportunity to -- to set a framework for A.I. regulation. You know, Trump has indicated that he wants to win the Nobel Prize. And if he can create a framework for regulating A.I., he might actually deserve it. He might actually get it. So, this is a rare opportunity that he has with Xi, who himself, Xi has also indicated a willingness to come to the table and talk about this.

KEILAR: So, besides shutting it all down, you write that there are three other potential approaches to regulating A.I. Take an air crash investigator approach, monitor the situation or flip the kill switch.

[15:10:05]

Walk us through each option briefly as you see it. What do you think?

WITT: Right, right. So,, it's all from aviation, right? When -- when a plane crashes, you know, what happens? Well, we send a bunch of investigators to the plane crash site. We interview everybody and we -- we create a really long report that future aviation officials can use to regulate aircraft and make them more safe. This is really effective.

Like airlines are really safe these days. We haven't had a major plane crash, you know, we almost never have them anymore. And that's partly because of the work that's done by the air traffic safety investigators. So, we should do that for AI.

When A.I. goes haywire, when A.I. breaks out of its container, when it causes damage in the real world, we should have an incident report that's commissioned by the government. The second is, again, just like aviation, we know where every plane is in the sky, every commercial aircraft. It's all publicly available information. It's all tracked according to some kind of air traffic control database.

You know, we can do the same thing with A.I. We can look at where all the powerful A.I.s are located, which data center runs they're doing, and kind of how they're being trained. And that visibility can give us a lot more eyes on the problem.

Right now, it's extremely opaque. It's very hard to know what OpenAI or Anthropic or indeed any of the Chinese companies are doing. But with a global framework like air traffic control, we can finally do that.

And I think the last piece, and this is something that is actually bipartisan, is that we should have a kill switch for A.I. So, the DHS in the proposed bill, the Department of Homeland Security, should be able to call Anthropic up on the phone if they see something worrisome and say, hey, you guys got to stop this run. You got to stop this A.I. It's out of control. Press the kill button. It basically hard stops the data center. It stops the A.I. from running. And then we can investigate what happened later.

So, I would say that we should slow down until all of this infrastructure is in place. And I will say it's not just me saying that. I'm hearing this a lot from A.I. scientists. I'm hearing this a lot from people who have worked at OpenAI, who work at Anthropic, who work for Elon Musk.

Like, if it was just me saying this, I think it'd be easy to ignore. I'm just some armchair critic. I'm just a journalist. Many of the leading technologists in the world want these things in place. The air traffic crash investigation proposal came from the single most cited computer scientist alive. So, it's really coming from sort of inside the house, if you -- if you understand what I mean.

KEILAR: Stephen, it is great to speak with you. Really appreciate the insights. Thank you so much.

WITT: Thank you. It was great being here.

KEILAR: And still to come, attorneys for John Brennan say the former CIA director has been subpoenaed in the investigation into President Trump's purported political enemies.

Plus, one of the Air Force officers who was shot down over Iran is speaking out about his harrowing escape and it is harrowing indeed. We have the details.

Then, later, a disturbing new report says migrants were held in small metal cages at Florida's detention center known as Alligator Alcatraz. We have that and much more coming up on CNN NEWS CENTRAL.

(COMMERCIAL BREAK)

[15:17:26]

KEILAR: Now to disturbing claims in a new federal watchdog report. The report says nearly 80 immigrants who were held at the so-called Alligator Alcatraz detention center were sometimes held in small metal cages. In some cases, that confinement lasted up to two hours.

The 34-page report reveals more details about the conditions at the Everglades facility before it was shut down in June. CNN's Priscilla Alvarez is following this -- this report for us. What else did it say?

PRISCILLA ALVAREZ, CNN CORRESPONDENT: Well, we've known that there were concerning conditions at Alligator Alcatraz over the year that it had been open. But now we're understanding a little bit more about the scope of those conditions from the Department of Homeland Security Inspector General. And perhaps most revealing in that is these small metal enclosures.

That's how the IG report refers to them. They were in -- you see a photo there of one that was outdoors. They said that there were at least two of those. It looks a bit like a phone booth. And then there were others that were indoors. And we understand there were 79 detainees who were placed in these enclosures for a few minutes up to two hours, the average time they found to be just less than an hour.

Now, the way the facility staff described this, because they talked to the inspectors during their January visit, was that it was, quote, calming areas. Or in other cases, they said detainees asked to be placed in those.

But the IG said something, and I've read a lot of these reports, that was pretty stunning, which is that they have never seen something like this before. And they have visited many detention facilities because that is their job to conduct oversight.

Let me read you one quote from the report. It says, quote, "The use of such restrictive spaces is highly unconventional and does not align with standards for humane treatment."

Now, we also asked the state of Florida about this because there was a bit of a tussle between who exactly was managing this facility. Was it DHS or was it Florida? In the inspector general report, they say DHS told them that it was Florida maintaining daily operations.

So, in their statement, they said, quote, "The left-wing media wants to focus on a confinement area for violent criminal aliens who have proven they cannot remain peaceful while awaiting deportation. While it adheres to all federal standards for criminal confinement, we'd make it even smaller if we could." I will say it is not the media focusing on this. It was the inspector general report that focused on these enclosures. And it was unclear to the inspector general, according to this report, what federal standards, if any, they were adhering to. They could not get a good answer on that. They lay that out in this report.

[15:20:06]

Other conditions that were of concern was the cramping. So, there were more detainees in a housing unit than there should have been. There were eight housing units that they had looked at and that there was also issues with access to clean water.

Now, some of that we had known from detainees who shared that with us, as well as their families and Democratic lawmakers. But this, again, showed the concern around Alligator Alcatraz, which, of course, closed in mid-June, but was a facility that was quickly erected and one that the Trump administration had touted at the time, and President Trump himself had even visited before those operations got underway.

KEILAR: Yes, it got so much attention, in part because of that.

Priscilla, thank you for the reporting.

New York and 21 other states are suing the Trump administration over a new rule that they say would make it harder for some immigrants to get a green card. The rule change, set to take effect Friday, would likely make it tougher for many immigrants to obtain the permanent resident cards if they use or are deemed likely to need public benefits like food stamps, Medicaid, or housing vouchers.

The states argue they would lose billions of dollars in federal funding if immigrants, particularly mixed status families, disenroll from programs due to fears about immigration consequences. CNN has reached out to DHS for comment.

And ahead, one of the Air Force officers shot down over Iran is speaking out for the first time about his harrowing escape and rescue by special forces. We'll have his story next.

(COMMERCIAL BREAK)

[15:26:15]

KEILAR: One of the two U.S. Air Force aviators shot down over Iran in April is speaking out for the first time. The officer identified only by his mission call sign, "Bravo," shared new details with CBS' 60 MINUTES, including the harrowing moments that he and his crewmate ejected from their F-15 jet. He also talked about how he was able to go undetected with Iranian forces on his heels and his dramatic rescue 50 hours after it all began.

Watch as he recounts the moment that he realized there was a problem with his parachute after his plane was hit.

(BEGIN VIDEO CLIP) BRAVO: My parachute was damaged in the -- the initial attack at one point when I looked up and saw no parachute. That was the most -- it was the most terrifying thing I've ever seen. So, I paused right then and I prayed. Lord, your will be done. But if I'm getting through this, I need help.

(END VIDEO CLIP)

KEILAR: CNN's Zach Cohen is with us now. I mean, it's an incredible account. And it sounds like, you know, I think you're supposed to hit the ground at like 10 miles per hour. He hit it. It must have been multiples of that, they estimate.

ZACHARY COHEN, CNN SENIOR NATIONAL SECURITY REPORTER: Yes, he says he hit the ground at about 70 miles an hour in this interview with 60 MINUTES. And it really is a remarkable story. You know, he says he was badly hurt after he effectively fell out of the sky and hit the ground going that fast. And, you know, he had injuries like a broken back, a broken shoulder, a broken arm, but still was able to pull himself off the ground and move. And not just move, but moving, hiking up a 7,000- foot ridge in order to avoid capture by Iranian forces who were actively searching for him at that time.

It was -- once he climbed up that ridge that he was able to meet -- send his first message back to the U.S. military to let the -- let them know that he was alive. But again, the question of how does one find the motivation to -- with a broken back to hike up a 7,000 foot ridge in that moment? Well, he actually addressed that during the interview as well. Take a listen to what he said.

(BEGIN VIDEO CLIP)

BRAVO: I had injuries, but we all trained to adapt and overcome the things that are -- that we're faced with. I moved as quickly as I could, despite the injuries, away from the place that I -- I landed. Never let a lack of motivation put you on Iranian TV.

NORAH O'DONNELL, HOST, CBS NEWS "60 MINUTES": You were thinking that?

BRAVO: Absolutely.

(END VIDEO CLIP)

COHEN: So the airman Bravo goes on to talk about how, yes, this is about his personal perseverance, his ability to, you know, outlast or stay on the ground in Iran for about 50 hours before he was rescued. But it's also more so a story about the teamwork over a hundred or nearly a hundred U.S. special operators were on the ground inside of Iran searching for this downed airman and his fellow pilot who was rescued beforehand. There was hundreds of other aircraft and support staff. U.S. intelligence officers were also actively conducting a -- trying to distract the Iranians, essentially in a very intentional way to try to buy more time for these special operators to find "Bravo" and "Alpha."

And look, I think that this speaks to, again, Dan Caine, who's interviewed in that 60 MINUTES piece. And Admiral Cooper both made clear that they're willing to send in more U.S. troops if needed to find Bravo. And fortunately, they were able to locate him within 50 hours and he was alive.

KEILAR: It's interesting that Dan Caine and others sat for interviews for this. I know that multiple sources told you and your other fellow reporters who reported this out this weekend that there was an internal push for them, these officers, to sit down for this interview. What did you learn?

[15:30:05]