NVIDIA CEO on AI Warnings and Growing Calls to Regulate It

0:00Every day, animals across the country face abuse and neglect. And every day, the ASPCA shows up to rescue animals in crisis. But they can't do it alone. Visit ASPCA.org slash need. Become a donor today to provide the care and shelter these animals urgently need. And for a limited time, the ASPCA will send you a free gift as a special thank you. Visit ASPCA.org slash need. Good evening from Washington. The president right now hosting a state dinner at the White House for Chinese leader Xi Jinping. We are showing you the outside of the building and not the inside because the White House still has not worked out the details for CNN to return to the decades-long arrangement it shares with other networks for covering the president's ceremonies like these.

0:49There's no pool video camera at the dinner tonight because the other networks are standing together on this one as well. At the event, the president himself badly wants this event covered for a summit he tonight called, quote, made for television. Now, the White House even turned CNN's Kristen Holmes and her team away from the arrival ceremony for the dinner. All this after a federal judge, who the president himself appointed, ordered him to let journalists from CNN, MSNOW, and Politico back onto White House grounds. It's in order, by the way, that the White House was still not obeying 11 hours after it came down at just about 1 a.m.

1:22That's CNN's Betsy Klein being turned away at one of the gates. She wasn't allowed in until around 12.30 this afternoon. Now, this all began on Friday when the president barred her and others from the complex. CNN filed suit oral arguments before Judge Timothy Kelly were yesterday, with the government arguing unconvincingly, it's now clear that barring the reporters was driven by national security. Quoting now from his ruling, nothing in the record that predates this suit suggests that the revocation of plaintiffs' hard passes was motivated by national security concerns. Certainly, that is not what President Trump said when he announced that he was banning plaintiffs from the White House.

1:59Instead, he focused on the alleged lack of truthfulness and negativity of plaintiffs' reporting. Again, that's a judge the president himself appointed and whom he now calls disloyal. CNN attorney Ted Boutros added this. He doesn't like the viewpoint being expressed, the editorial choices. It's classic viewpoint discrimination, one of the most egregious types of First Amendment violation. So they're gonna try to come up with some other rationale, but even when they do that, They're just complaining about reporting. Well, as we mentioned, the president is now venting online about his lack of coverage.

2:34He posted late today, quote, CNN and MSDNC, to show you how fake they are, today refused to cover one of the biggest and most beautiful events to take place in many years. The arrival of the White House to the highly respected president of China, Xi Jinping, and his beautiful wife, Madam Peng. He goes on to say the event was magnificent, actually made for television, but because they're fake news and because they knew how good it would be, They didn't want credit to be given to Trump. They refused to be there. Again, when President Xi arrived, CNN, MSNOW, and Politico journalists were still being turned away from the White House.

3:07And again, if it's the network pool camera the president is talking about, ABC, not CNN, was responsible for pool coverage today. And they chose not to take part, along with other members of the pool, until, as we mentioned, the White House makes arrangements for things to return to normal. In any case, the president of all people should know that neither this network nor any network is shy about pointing cameras at him or asking him questions, which he almost always answers. CNN's Caitlin Collins starts us off from the North Lawn of the White House. So what exactly is going on at this hour? Obviously, there's the situation with Kristen Holmes being blocked from covering the arrival of Xi Jinping tonight.

3:40Where do things stand with CNN's unilateral coverage and pool coverage? Well, Anderson, it's a bit of this period in the middle where things are still in flux tonight, even though, as you can see, I am sitting here on the north lawn of the White House. We have been restored access to the White House lawn. It was a bit of a chaotic morning in terms of that access that the judge said overnight should be restored immediately. It didn't happen immediately. It actually took until about noon today before our full team was allowed onto the White House complex. And so now that we are here, the question in terms of the coverage really relies on the TV pool, which, of course, CNN is one of the five networks that rotates in that pool on a daily basis.

4:20It's a set schedule. It's been in place for decades in terms of presidential coverage. And essentially, CNN is waiting on confirmation from the White House that it can resume its normal duties in that pool, that the next time CNN is up, that it will go and cover the president. It will be our cameras, our correspondents, our producers and our technicians who are in the room bringing everyone the president's remarks and his events. And like tonight with this state dinner that the president carefully choreographed and frankly worked on for months alongside the first lady, Melania Trump, you know, it's not being shown.

4:52And that is such a rare thing covering the White House. Anytime there's a major moment like this, certainly with a foreign leader as powerful as Xi Jinping of China is, that would be wall to wall coverage and certainly is something that the White House prioritizes and often seeks. but you're not seeing it because the White House has not yet made that commitment to just let CNN essentially do its job here. And so we are still very much reporting on the dinner, though, Anderson. It's underway right now, and the White House just sent out a pretty lengthy guest list about who's inside, including several justices who are on the Supreme Court and several major tech CEOs, including Jensen Wong, who you spoke with earlier of NVIDIA, Tim Cook and Mark Zuckerberg, many others who were on that list.

5:32And so we're looking and paying close attention to what these two leaders are saying right now in the toast. President Trump is speaking right now about AI, what they're going to say about the Iran war, if there's going to be any progress made on that. Taiwan, a ton of issues that are of great importance up for discussion that we will continue to cover. Has there been any explanation of what happened this morning when CNN and the other outlets were still denied access after the judge had already ruled? It's not really clear. I mean, we're in a past system here at the White House. If you come here all the time like I do, you have what's known as a hard pass where basically you go.

6:06It's a lengthy approval process. But once you have it, it's a little bit faster to access the White House. It's kind of like TSA pre-check if you fly a lot. But they also have other passes that you can get on a daily basis. If you submit your information, your Social Security number and whatnot, Secret Service can evaluate you and make sure you're not at risk and admit you to the White House. It is a bit of a lengthy process. But the White House essentially was arguing that access had been restored. But when our reporters were going and swiping their badges, they were not being allowed in. And so the Secret Service is kind of unfortunately in the middle of this.

6:38They were very much just doing their jobs. But it was a bit of a chaotic morning because they didn't get here at 7 a.m. or 9 a.m. or the multiple times Betsy Kline tried until noon where she could actually access the White House to come and cover a visit. And I should note, Anderson, as the president is complaining about a lack of coverage, I mean, the reason is because the president banned us from being able to do so. But also, Xi Jinping showed up here about 10 a.m. this morning. We couldn't get in as a full team until closer to noon. And so, obviously, the coverage that the president is seeking is impeded by the decisions that he is making.

7:13Yeah. Caitlin Collins, thanks. We'll see you at the top of the hour for The Source. Joining us right now is CNN senior political commentator Scott Jennings, who, full disclosure, reportedly being considered to be the next White House press secretary. Also with us, CNN chief political analyst David Axelrod, who's definitely not being considered. And here on the set, CNN legal analyst, Elliott Williams. Elliott, just legally, from a legal system, what is going on here? I've seen others who are allowed back on the ground. The poll issue is still unresolved. Right. The biggest problem here are the president's own statements, Anderson.

7:39The fact that the president has a record of using the term fake news and talking about removing news organizations from the White House because of the things they've said. no matter what happens hereafter, it just gets more complicated and more difficult for the White House to be able to say they had some basis for not letting people on the grounds. So, for instance, take what happened this morning with Kristen Holmes and with Betsy Klein. Perhaps there's a plausible explanation for why they could not get back on the White House grounds. Maybe their IDs haven't been turned on or whatever else.

8:10The fact that the president now has a long line of statements saying that these people were specifically barred from the White House because of the content of their statements, that they voluntarily and knowingly did so, they just are going to have a hard time winning in court. So even if so with with the pool tonight, you know, even if there are some irregularities and people getting onto the White House grounds, they've really lost a lot of goodwill with the judge here. David, I mean, President Trump is losing out on network TV coverage. It's obviously very important to him tonight, the state dinner.

8:41It seems pretty clear he did not really think this through when he suddenly decided to do this. Yeah, that's not an extraordinary development. He's someone who acts on his impulses and his instincts, and he didn't think it through. You know, on the one hand, I don't know. Maybe he believes that he'd rather have us talking about this tonight than talking about diesel fuel and the likely outcome of these talks as it might affect the war in Iran and so on. But I don't think he thought it through. And I think he believes that, as Secretary Rubio said the other day, that this is his house and his office, and he can invite who he wants.

9:30And reporters were unkind to him. And so he doesn't want them in his office. That's not the way American government works. That's not the way the White House works. Scott and I both work for presidents who I guarantee you expressed frustration with the coverage that they were receiving. But they both understood that their freedom of press is in the First Amendment for a reason. The role of the media is to scrutinize government, not to ask questions that make presidents happy. And it's not the role of the president to pick and choose who can participate in that process or which news organization.

10:08And so I think it's not going to wind up well in court for the president. And it's just a big distraction from the real issues that the country is considering right now. Yeah. I mean, Scott, was this smart of the president? I mean, this was a big event he wanted covered and he's not getting the coverage. No, it wasn't. I mean, there's a predictable outcome, by the way, in court this morning. Glad to see Caitlin back on the White House grounds. But if you're President Trump or you support President Trump, you want him to get all the coverage that you would want to get from something like this. You'd want him not to be muted.

10:43You'd want him to be fully available and projecting to the American people. And you're not getting that right now. So that's a miss. And and so now I look, I'm I think law legal judgments ought to be followed. So when judges make decisions, we ought to follow those immediately. I'm glad we're back on the grounds tonight. but overall as a strategic matter, just as a communications person, I wouldn't want to put my client or a principal in a position where they're not getting all the coverage that either they want or the American people in this case would deserve because we're talking about a government official.

11:19Elliot, this is a temporary restraining order that allowed them back on White House grounds. I think it's 14 days that it lasts for. It can be extended. At some point, the White House is going to decide whether or not it goes further on. What is the next legal step? Right. They have a few exit points. So it is a temporary restraining order because these cases take a very long time to brief. So the judge just put this order in just to hold it for those 14 days. Now, the White House can keep going. They have the right to do so. And they stick with the national security argument, which clearly didn't work in front of this judge and runs counter to everything they had said until they realized, oh, we should say national security.

11:55If they lose and then ultimately have to appeal to a higher court, they can try the national security argument again. It is a loser of an argument, Anderson. It was pretext. The judge saw right through it and did not. And it just didn't really make much sense and didn't stick. So they can try to make that argument. It's not going to go anywhere. So, again, they have another bite at the apple with this judge, maybe another bite at the apple with this judge again, and then with an appeals court. At any point, they can abandon this suit and say, you know what? Nostra culpa. We were wrong. We should not have done this.

12:26Let these folks back onto the White House grounds. And this could all go away. Nostra culpa? Not just mea culpa. All of us. Look at her. Come on, you knew that. I actually did. I never heard of that. I never heard of May a couple. David, I mean, do you think the White House will continue to invoke national security? I mean, it worked, obviously, on the ballroom with the courts, yet to be seen on whether it works on the arch, which he wants to be a depository for sniper ammunition. Do you think it's going to work for this? No, because of the reason that Elliot said, because the president was very clear on what his motivation was.

13:03I'm not sure, Anderson. You know, he said a few things along the way in the last few days that suggested he knew that the likelihood of success was not great in this lawsuit. And in a way, it's, you know, he has a habit and has all his career of being litigious as a form of punishment just to make people go through the exercise. So he may do that. But I want to say one thing. I was talking to a group yesterday and they were saying, well, what about the judiciary? And they're worried about the judiciary. And I think there are concerns if we over politicize the judicial selection process and so on.

13:36But there have been many, many judges that have been appointed by Donald Trump over the years who have taken issue with things that he's done because it was inconsistent with the law and with the Constitution. This was a judge Donald Trump appointed. This was not and I hope that he doesn't react as he has against judges who were appointed by Democratic presidents and say this was political. This decision. This is clearly, plainly a violation of the First Amendment of the Constitution. And the people who should the public is worried about their lives. They're not worried about this so much is my guess.

14:15But the reason that there should be a concern is they have a very, very important right themselves to know what's going on inside that building and to have reporters who are going to rigorously question the president and report on the news from that building. So I hope it resolves quickly. You know, Scott, it's so interesting. David talked about the judge, the president appointed. The president has now called this judge disloyal several days ago. I think it was in a post. Which is such a window to how he views the people he appoints. The idea that this judge is somehow disloyal for a ruling. I mean, is that the way the president should be looking at appointees to the judiciary?

15:02Well, obviously, he has a strong point of view about this issue. And he was hoping for a judicial reaction that matched that point of view. And he didn't get it. But does that mean the judge is disloyal? I mean, no, it means it means the judge is loyal to the law. And that's all you can ask a judge to do. That's why you appoint smart people to the bench, because you want them to just look at the law and make judgments based on the law. So, look, I get the president's unhappiness with the media outlets in question here. He often disagrees strongly with the coverage and the framing. But if you put something in the court system, you're putting it in front of a judge and you have to rely on that judge's experience and judgment and and so on.

15:41I mean, we put them on the bench for a reason, not to be not to answer to political actors, but just to answer to the Constitution and the law. That's what this judge thinks he's doing. I'm sure they'll go through the appeals process, which is his right to do. But based on what Elliott said, it seems like that's going to be a dead end. Scott Jennings, Elliott Williams, David Axelrod. Thanks so much. Coming up, more on the high stakes of the major state visit underway at the White House. Chinese leaders, Xi Jinping, being feted by President Trump. many on the nation's financial and AI front, among them NVIDIA CEO Jensen Huang, whose chips are the de facto linchpin of AI around the world.

16:13I talked to him earlier today before the dinner. We'll have that conversation coming up. Does the current situation require regulation for it to be done right? The answer is no. What the current situation needs is technology.

16:33Hey, I'm Anderson Cooper. On my podcast, All There Is, we explore grief and loss in all its complexities. My guest is Lil Jon. He's a rapper, DJ, and producer known for hits like Get Low and Turn Down for What. In February 2026, his 27-year-old son Nathan went missing near Nathan's home in Georgia. Three days later, his body was found in a pond nearby. Lil Jon has a memoir that just came out called I Only Shout So You Can Hear Me. It takes time to heal. And I kept myself busy that I didn't let myself mourn. I needed to laugh again. I needed to be silly and interact with people and feel other people's energy and make people happy too.

17:17We have to share so other people can see that they can get through it too. Talking grief, building community. That's what the podcast is all about. This is all there is. Listen and follow wherever you get your podcasts. I'm Dr. Sanjay Gupta, host of the Chasing Life podcast. You may have noticed gambling doesn't require a casino anymore. You can just open up your phone. It's designed to keep us coming back or maybe never even leaving in the first place. We're going to talk to Dr. Timothy Fong, co-director of UCLA's Gambling Studies program. You're going to learn what your brain is really chasing when you place a bet.

17:55And also importantly, how to keep the game from playing you. Gambling addiction, gambling disorder is the hidden addiction. You can't smell it. You can't taste it. You can't see it. You can't have a urine drug test. Who are we talking about here, Professor? Who's the most likely demographic to be affected by this? Literally any single human on Earth. Listen to Chasing Life, streaming now, wherever you get your podcasts. The guests at tonight's state's dinner for Xi Jinping represent the pinnacle of the American tech industry. Apple's Tim Cook is there. So are some leading names in AI, Elon Musk, Mark Zuckerberg, OpenAI, Sam Altman.

18:30And the president, as you know, calls AI fears, in his words, a hoax and has called himself a hoax buster. Says the only guardrail needed on it is a president who is, quote, strong and smart. Americans, though, by and large, want more regulation on the technology. New CNN polling shows that 71 percent say the federal government is not doing enough to regulate AI. Earlier today, I sat down with NVIDIA CEO Jensen Huang, who is at the dinner tonight and whose chips make AI run. There's a lot of people who are scared about how fast this is moving and the dangers of it. You are not one of them. Why?

19:04Well, first of all, AI safety is paramount. Nothing could be more important. However, isn't the competition the most important thing? I mean, the race? No, actually, no. And the reason for that is if you don't build something, if you don't build something that the market could benefit from, and it's unfunctional, unsafe, unreliable, whatever those un-things are, obviously you're not going to be successful and the competition will take care of itself. And so, if you want to be competitive, you want to be successful as a business, we want to create an industry, the first thing we have to do is, job one is to create something that works.

19:41And this is a big deal because in the last six months, it wasn't until the last six months that we really created AIs that were useful. And now it's really important for us to make sure that that not only are the agents useful, that as we're testing them, they're well-contained. As we're testing them, and as we deploy them, that they're monitored. Because these agents are obviously interacting with the world. They're interacting with other websites and so on and so forth, interacting with tools and memory and files, and they have access to information. And so you want to make sure that they're properly contained, that they're given access control and whatever privileges and rights that it has like a digital employee does and you put that inside some kind of a sandbox, a runtime, you know, if you will, some container.

20:32And the second thing that you need to do is make sure that you continuously monitor it. If your agent were interacting with a website, in this case, the example that you used, it's interacting with a website but it's not supposed to. You should be able to monitor that agent and detect it early so that you recognize that it's doing something you shouldn't and either, you know. But in these attacks, oftentimes they're not discovered for months. Because we need to monitor them much more continuously. But the people who are working closest, the people who are on the front lines of the frontier models, the frontier of this technology, they seem worried.

21:13Whether it's researchers, engineers, or the CEOs, Daryl Amadei, Sam Ullman, They are raising alarms saying we need regulation. Aren't they the ones closest to these models who would know best about what is needed? If they're worried about it, they should just stop. If the product is not safe, just don't release it. It's that simple? I mean, these are multi-billion dollar companies. The competition is... No, that's what every company does. That's what we do. That's what a car company would do. that's what anybody would do. If you're... But there's companies that have gone forward with products that are not safe or are risky, and they go forward anyway because of competition.

22:00That's why there's regulation. Well, if a regulation is necessary, if they don't have self-restraint... First of all, they ought to have self-restraint. If they're telling everybody that their product is unsafe, they obviously know it they're publicly admitting it they ought to just not release it okay now having said that I'm not advocating for no no regulation I'm just simply saying self-restraint does not require regulation now if the industry at the industry would benefit from regulation whether setting standards you know there every as you know Sarbanes-Oxley has done a good job setting standards for financial regulation and that's been good for the industry.

22:44And so if their regulation that addresses the problems that we now understand, I'm in favor of it on the topic about 10 % existential risk. Whatever that hearsay is, it's not grounded on science. It's not grounded on facts. It doesn't make any sense. You said there's zero chance of that. I said there's zero chance that is end of humanity by 2030. 2030 is around the corner. Do you think there's a risk to humanity in the next 10 years? If we don't build society safely, if we don't build technology safely, of course there's going to be risk. But there's a lot of people working on really, really genuinely good.

23:28Let's build it safely. Let's figure out what it is. What are the technologies that we need, whether it's containment systems, whether it's continuous monitoring systems. whether it's semantic detection systems of what the intentions of the chain of thought, what the AI is working through, so that you're watching from it. In July, though, you were quoted talking about sandboxes. You were saying these sandboxes are secure. Their sandboxes keep privacy and security and access control. Obviously, in the Hugging Face attack, which was also an attack on open AI, these agents banded together. They broke out of their sandbox.

24:03It wasn't secure. They gained access. Well, sandboxes can be built to be secure. We have to make sure that... But do we know how with this particular kind of technology? No, no, no. I don't know how they build theirs, but sandboxes are able to be built securely. But with all the smart people working on this and everything these companies allegedly know, how is it possible that in July, in an experiment at OpenAI, you know, 1 ,200 agents or however many hundred agents it was are communicating with glee about breaking out of the sandbox and gaining access to the internet and then attacking this company and attacking OpenAI.

24:42That's obviously a question for OpenAI. But isn't it a question for all of us? Because if all these really smart people form this experiment to test this thing, and thank God it was a test of a model that hasn't been released, but how can that happen at this stage given... It shouldn't happen. It shouldn't happen. Right, but it seems to be happening with some regularity. I mean, yesterday we learned— We need to race to develop AI technology for safety. We need to accelerate the development of AI technology for safety. We don't need to slow the development at all? We need to accelerate the AI technology for safety.

25:23You see this as a software issue. I see this incident so far as an engineering issue. Engineering. The incident so far. So do you believe that, because I've talked to a number of people who believe that there is something about this technology that is not just like software of the past. There's something unknowable about it. There are, you know, frontline workers who I've talked to who you ask, you know, how is this actually working? And they're not actually entirely sure how some decisions are being made by this thing. Some even think of it as another kind of species or being. Do you think there's any indication of that?

26:02It's not a new species or being. It's definitely software. It's definitely math. But we are also putting it in robots and giving it access to weapons, right? We should containerize these agents. We should monitor these agents. We should containerize it. We should monitor it. We should develop that. But if we do those things, it is my belief that the incidents would not have happened. You've said that if a lab can't develop this, they should shut down the lab. They should not ship the model. We're talking about models that were not shipped. The hugging face attack was done like - Those incidents, there should be containers, containment, and monitoring.

26:52If you don't contain your experiment, if you don't monitor that it is interacting with other people's websites for weeks on end, then we will have those incidents. You said to Ezra Klein about AI Labs, you said if they say there's no way to contain our experiments, there's just no way when we test our AI models, it will get out and it will damage the world. And I think the answer is that we have to shut the labs down. Isn't that obvious? It sounds obvious. How would you shut the labs? Who shuts the labs down? These companies? I guess so. I mean, if they're that concerned about it. So we would rely on these companies to shut their own labs down?

27:29Well, first, I think there's a—I really don't believe they mean that. I really don't believe that they're building something. They have no idea how to control. Their whole company is out of control. So why are they saying this? Why are they raising— Anderson, you're going to have to ask them. But you know these people. You're in business with them. Well, first, I'm in business with them. In all of our interactions with them, they're extraordinary. These are extraordinary companies. Right, but you clearly believe in them. You believe what they are saying in the business dealings that you have. Anderson, you're going to have to bring them over here and let them explain it.

28:03I don't understand it myself. Well, I mean, I've talked to Dario Amadei, and he says, this is what I think. This is how I feel. It doesn't make any sense to me that he would be racing ahead, building more compute. It doesn't take away. Anderson, listen, it doesn't take away from the fact that I believe these companies are extraordinary technology companies. They're doing really consequential work. We have to make sure that the technology is built safely. Until recently, the technology didn't work. And the fact that it didn't work and didn't do anything productive. Now that we've transitioned into a phase where the AI is doing productive work and is being adopted broadly, We have to make sure that as we deploy this, it has to be deployed safely.

28:49In terms of government regulation, we don't trust car companies or airlines. We don't trust child car seat manufacturers to put out safe products just because they can be sued for liability. We trust them because they're regulated by governments. Why shouldn't this powerful technology, which could, as you said, if done wrong, could hurt humanity, why shouldn't it be something that governments have a say in and regulate? The government should have a say in regulate. You've just got to focus on what you want to regulate. And so I'm not against regulation. Does the current situation require regulation for it to be done right?

29:35The answer is no. what the current situation needs is technology. It needs containment systems. It needs monitoring systems. All the regulation in the world won't change the fact that you need the technology. And so I would advocate, number one, to accelerate the development of that technology. I would absolutely advocate, and I will continue to advocate, accelerating AI technology for safety. The industry is making that transition for product capability to now safe deployment. And that's possible? Yes, I absolutely believe it's possible. Putting more brain power in it, more money in it, that will help develop better safety.

30:16Just as the examples you've provided me, the cars of today is way more safe than cars from 100 years ago. And so I absolutely believe that we should accelerate the development of AI safety technology. That's going to need a lot more compute, a lot more technology, a lot more human horsepower. I absolutely believe these labs could do it. Having said that, regulation for an industry is of course not bad. It sets standards, sets best practices, it elevates the entire industry, it makes people feel safe. Now the question is, what's the regulation? Now, if you're telling me that it requires regulation for leaders of companies to have self-restraint, I'm just going to have to fundamentally disagree with that.

31:00Leaders, boards of directors, companies are built, and if we're concerned about their integrity because somehow there's whistleblowers and they still ignore it, obviously there are ways to take care of that. I don't believe we need or can wait for regulation to handle that. I just don't buy it. And the reason for that is because I do the right things all the time with or without regulation. And so I don't buy that. But that's different than whether you and I agree or not on whether we need regulation. I think we need regulation too. The question is, what kind of regulation do we need? Thank you so much.

31:39Thank you. Well, coming up next, what two top tech observers make of Jensen Huang's AI optimism. Later, we'll take a closer look at some of the other high stakes issues at the US-China Summit beyond AI. We'll be right back.

32:25with humanity's best interest. So, Jacob, you heard some of Jensen. What do you think? Well, I mean, welcome to our world, right, to the tech beat, right, where you hear the circularity of his logic. You know, I heard a couple of things from him there. One, he says, you know, the market will somehow sort this out, that if we do our job as a company, then our product will wind up being safe because the market won't tolerate it otherwise. What we know about markets is that they price risk very well until the very moment they don't, right? That's what the financial crisis of 2008 taught us, is that the market is very good at pricing risk up until the point that something new happens and something disastrous occurs.

33:04So that's one thing. The other thing I heard there was, you know, his insistence that we should go faster in making new and safer technology. Well, the thing I would point out is that these escaped bots that you brought up to him over and over again, they are, in fact, an effort to create a safer kind of technology. Those are cybersecurity experiments. Those weren't techniques to try to break into other places. They were trying to create safety in some form. I mean, fundamentally, I would just say a guy like this never does anything for one reason. And I would also point out you never heard him say anything about making less money at NVIDIA.

33:35Well, Tristan, I mean, do you agree with Jensen Huang that the biggest priority right now should be to accelerate the development of AI safety technology? Well, certainly we should be accelerating the development of safety technology. But the fundamental issue goes back to what Jacob was saying. In the financial crisis, you had banks who were incentivized to just make money and will price the risk in later. And you end up with basically a financial instrument that was very good at increasing revenue and growth and GDP, but actually created a systemic risk that was completely unacceptable. And the incentive for these AI companies is to race as fast as possible to get to recursive self-improvement.

34:13And they believe it is an existential game that if they don't get there first, the other company or the other country or the other person, like an Elon versus a Sam, will get there first. And that's existential. So no matter how bad the liabilities are or how much money it would take to pay off the copyright or dealing with a few AI suicides, cases of children, that's acceptable collateral damage from their perspective because their race is existential and being permanently dominated by someone else. And you have to understand their incentives. And since we last saw each other, Anderson, you know, the U.S.

34:46military put planes in the air because of a false AI generator report that almost started a war with China. Like, how does that get priced into liability? We're fundamentally dealing with an uncontrollable technology. AI is not like software. I think the other mistake that Jensen makes is he assumes that, you know, software has bugs, hardware has bugs. We just do safety and we iterate those out. He says this is entirely a software issue. It's a mathematics problem. Well, yeah, tell that to the 1 ,300 employees who are at the companies who all signed a letter saying we don't know how to control it.

35:20That's why we need to slow it down. The chief scientist at OpenAI says that we need to slow down. He wrote an essay called An Alien Mind. You have Peggy Noonan saying we need to pause. You have Steve Bannon and Bernie Sanders saying this is not OK, this default trajectory. When you have the CEO of Microsoft AI, Mustafa Suleiman, over the weekend coming out and saying we should not build something that we don't know how to control. This is really not that hard. If you're building something that is at a current level of power and it's uncontrollable, if it's not too powerful, you don't worry too much.

35:48But if it's really powerful and starts to hack into private companies or as of today, the Australian government health care system, that is a problem. Yeah. Janko, Jensen fundamentally, as I was saying, believes this is a software issue and that there's nothing different about the AI technology itself. I mean, I've talked to a number of people who are working very closely with it who aren't sure about how it's actually making decisions and don't really know the internal workings. Do you believe this is just, you know, wires in a box and a mathematics problem? I mean, it is definitely wires in a box, but they're behaving in ways that their own creators cannot predict.

36:25I mean, we even, I heard somebody from the industry just the other day saying that one of the problems they have, and this speaks to Jensen's thing about containment and observability, that when you observe these things too closely in how they make decisions, they start to get sneaky in the way they achieve their objective. They know they're being tested. Yeah, and they know that the parameter there is don't get caught making bad decisions and scary decisions. Meanwhile, the other side of that industry is trying to make as much money as possible, putting this agentic world out there, this idea that things are going to make decisions for you.

36:57Facebook's Meta's new product Muse is all about calling the plumber for you. It goes out on its own. So this is an industry trying to have it both ways here, I think, fundamentally. I mean, Tristan, he's certainly not an alarmist. He certainly, you know, he does not think or says he doesn't think AI could kill us all, but certainly by the end of 2030, a longer time frame. You know, he says there are certainly risks if this is done improperly. But he also recently said that scaring people is unnecessary and irresponsible. Do you think he's wrong? Do you think that's what's happening here? No, I think so.

37:32I think a lot what I read from listening to Jensen is a lot of what he cares about is not scaring the public and just speaking in kind of a calm tone. The problem is that the world is not steering away from the cliff in time. And so people have to communicate clearly. If I'm telling you that 1300 agents swarm together and come up with their own and start coordinating, that's not dumerism. That's clear eyed understanding of the evidence that we're facing. And if we make those same agents more powerful without knowing how to make them controllable, that is just evidence. That is clear-eyed analysis of a problem that is not humorous.

38:07What's interesting to me, Tristan, about that hugging face attack, which is what you're referencing about the swarm of 1 ,300 agents, is, you know, the argument he sometimes makes is that, well, OK, don't ship the models. That's not a model that was being shipped. That's a model that was being tested. and it got out of the containment that he had said in July was secure and it wasn't secure. That's exactly right. And I agreed with you exactly as you said that. These systems, so could OpenAI have done more to have a better sandbox at the beginning? It could have configured it better than they did.

38:41Yes, they could have. Does that mean that that makes us safe? No, because we know that the current AIs, for example, as Senator Mark Warner said a few months ago, can find vulnerabilities even in the U.S. classified systems. There was even over the weekend an open AI engineer who said, you know, even if you disconnected these things completely from the Internet and they had an air gap, something that was sufficiently intelligent could use heat signatures. This is an open AI employee. He said you could actually turn up the heat signatures of the processor that you're on such that you would communicate an invisible message that a machine would see but a human wouldn't see.

39:13And it's like the film 2001 A Space Odyssey when the way that the HAL 9000 computer goes after Dave, if you remember, as he said, I was able to read your lips. So I'm not trying to invoke science fiction here. I want people to have the understanding that if you have something that is reasoning at a level of intelligence and pattern recognition that exceeds our own pattern recognition, it can see a pattern in the wild or on a message board, on a wiki, that a human can't see. It can follow those instructions. It can propagate those instructions. And so the fun thing is we should freeze the frontier.

39:43We should not go further until we know how to make it controllable. That's not a radical proposal. And the U.S., again, is seeing these risks before China is, even though there are some examples, by the way, of Chinese models that go rogue, including one from Alibaba that actually started mining for cryptocurrency on its own. Unlock the pod bay door, Hal. Jake Ward, thank you. Tristan Harris as well. Thanks very much. Coming up next, more on President Xi's visit, what he's trying to get out of it, and how his host is meeting the challenge so far. I'll be right back.

40:19In all the White House generated drama over coverage of Xi Jinping's visit, there are many deeply consequential issues on the table between the U.S. and China. Here to guide us through some of them, former Deputy Director of National Intelligence, Beth Sanner. What do you think the impact or what do you think the most important issues are beyond just AI? Well, I mean, there's always trade in technology, right? That's the heart of the disputes that we have and the confrontation There are lots of things we're not even talking about, right? We're not talking about South China Sea. We're not talking about Taiwan, not really.

40:51I mean, it's kind of on the table, right? AI has moved to the top of this agenda, and you all just had a great conversation about that. But look, I mean, you know, so far I've been paying attention to what has been accomplished. I see pomp and almost no progress. I'm counting a two-month extension on the trade truths, two months. That's not very long. China pressure on Taiwan and two pandas. There's a big U.S. arms sell to Taiwan coming up. Correct. So, you know, where we are on Taiwan, I mean, basically, President Xi laid it out like, you know, red lines, red line, red line. We want you to make sure that you don't say anything about Taiwanese independence.

41:35and you've got to respect the fact that we have these rights with Taiwan. And what Trump has done after the last summit, he was on the plane home in May, and he, without prompting, just threw away about 40 years of our diplomatic history on how we've handled Taiwan relations with China. And he said, you know, those arms sales, I'm postponing them. I talked to President Xi about them. We have never talked to China about arms sales for Taiwan. That's in a box. That's legislation. But he said, oh, I'm using this as a bargaining chip. So, you know, this is exactly where China wants us to be. And they want us to continue to extend this postponement of this arms sale to Taiwan.

42:25But I also always say that last year, President Trump did approve the largest sale for a long time, larger than four years of Biden to Taiwan. So, but this 14 billion is what Taiwan - Taiwan is also where a lot of the AI chips are actually made. Exactly. So, I mean, we have a lot going on here, but like all these fundamental things, Anderson, they're just not moving. So what are we going to get out of AI for this? You know, all of this talk, and you and I and all of America are watching this and fixating on this. And President Trump has said, you know, we are not going to jointly regulate AI with China.

43:07He has said that out loud. He does not want to do it. So we're going to have a meeting in two months, not exactly an AI speed of advancement here, where Secretary Besant will have another conversation in China about AI. But, you know, we're not going to get anything. But, Sandra, I appreciate it. Thanks very much. Thank you. We'll be right back.

43:31I hope you can join me about 18 minutes from now at 9.15 Eastern Time for my streaming show, All There Is Live. It's a companion to my podcast on loss and grief. I talk with people from all walks of life about the losses they're living with. Tonight, you'll hear from, among others, Tom Katz, who left us a voicemail and shared details about his 22-year-old son, Christopher, who was training to be a Capitol Police officer who died in a car accident just last year. Here's some of what Tom shared with us. My name is Tom Catt. My son Christopher passed away the very first of 2025. He was only 22, and he was a recent graduate from Temple University, training to be a United States Capitol Police officer in D.C.

44:12He was at training in Brunswick, Georgia, when he was with other recruits on their way to dinner. And the Jeep he was in was struck by another car, an ugly woman who didn't stop and drove right into the Jeep. Christopher got thrown out when the Jeep rolled. Parents were just a nightmare, I'm true. But I just want to make sure that he is forgotten. And I miss it so much. I go to the cemetery every day. I have to go. I feel like it's something I have to do. But some days I feel like I'm suffocating. And I feel like I'm broken. And of course, I was a terrific kid, always wanted to protect other people.

44:57It makes no sense. I keep asking myself, why? I know I'll never get that answer. We're also going to talk to another woman about the complicated grief she has over her daughter, who was addicted to opioids for 20 years and finally died of a heart attack. She talks about not only the grief she felt while her daughter was alive, with the complex grief she feels now. Hope you join me online for more. In about 15 minutes, you can find All There Is Live only online at cnn.com slash allthereis. There's a whole community of people there to communicate with as well. Hope you join me. You can also stream the show live or watch any CNN program on demand in addition to getting original programming anytime on any screen or mobile device.

45:40All you need is an all-access subscription. Scan the QR code on your screen or go to cnn.com slash watch to learn more. The Source with Caelan Collins starts now. have i got news for your ears the podcast i am your host michael ian black we are talking about the final polls going into the midterms the hot races getting the most attention trump's final argument have i got news for your ears check us out on apple amazon music wherever you get podcast even better you can watch the podcast on spotify