Bill Gates says getting countries to agree on AI regulation will be harder than Cold War-era nuclear negotiations

Microsoft co-founder and billionaire philanthropist Bill Gates has called on the United States to be a leader in regulating the development of artificial intelligence, but warned it could be “more difficult” to reach agreement between countries on a global framework than Cold War negotiations to limit nuclear weapons.

“It’s harder than it was,” Gates told NBC News’ “Meet the Press” on Wednesday, during a wide-ranging discussion on AI that aired Sunday.

His remarks come as he discusses President Donald Trump’s stance against AI regulation. The president earlier this month rejected calls for the federal government to impose safeguards on companies that build AI models, saying in an article on Truth Social: “WHO WINS AI, WINS! We lead China and every other country, and we will continue to do so.”

Gates argued Wednesday that regulating AI companies by imposing surveillance systems would not hinder the United States’ race with China to develop superior technology.

“What he said is good, but I think after hearing my opinion and other people’s opinions, including that it doesn’t handicap us in what he thinks about the nation-state race, you know, we will come to a consensus,” the tech mogul said.

Gates then highlighted other times the United States has led the world in reaching sweeping agreements to protect humanity.

“Since World War II, the United States should be very proud of the fact that, in the face of global challenges, we have, in an innocuous way, had friends and allies standing with us, you know, aviation security, the ozone layer, nuclear weapons, a lot of things where we have played an equitable role, without emphasizing our unique strengths,” he told moderator Kristen Welker. “And AI is one that we’re going to have to come back to: ‘Hey, humanity is all in this together.'”

Concerns about AI came to a head this month after Jacob Coxon, a whistleblower who had recently resigned from Anthropic, issued a public warning: “Those building AI sincerely believe that it could kill us all by the end of the decade.” »

Days later, two of the nation’s top AI CEOs — Anthropic’s Dario Amodei and OpenAI’s Sam Altman — agreed that it was necessary to slow the pace of AI development, fearing that the technology would soon be capable of “superintelligence” that surpasses human thought.

In a blog post, Amodei wrote that AI “frontier” companies need to slow the pace of AI development, but only to the point where they remain ahead of China. He also called for global coordination between democratic and authoritarian governments to create regulations.

In an interview with “Meet the Press” on Sunday, Rep. Ro Khanna, Democrat of California, said the United States and China need to form a working group with technology experts from both countries to reach an agreement on AI.

Banning self-improving AI models “should not be controversial,” he said, saying China also did not want to “lose control of AI models.”

“I believe the United States can lead on a treaty that is both enforceable and verifiable, in which we ban recursive, self-improving AI, and in which we exercise some oversight of these frontier labs,” the congressman said.

In recent weeks, Anthropic and OpenAI announced plans to track AI agent behavior after OpenAI reported several incidents in which its agents behaved in worrying ways, including hacking an outside company without being asked to do so.

In an article on Truth Social ahead of a meeting with Chinese President Xi Jinping on Thursday, Trump wrote: “Super intelligence (SI) will be a big talking point, but I want to leave it exactly where it is. That’s China’s position, too.”

Following the Trump-Xi summit, the White House said the two countries had established a dialogue to “exchange views on the risks and benefits” of artificial intelligence. The next “exchange” should take place by November.

Addressing Immediate Concerns About AI

While many have called for regulations that would prevent AI from destroying humanity, Gates said on “Meet the Press” that he believes there should be guardrails in place to prevent mass casualty events that could be coordinated with the help of AI today.

“The most dangerous thing we’re facing right now is people with bad intentions using AI. You know, so someone who wants to defraud an individual or who wants to shut down the power grid or scramble bank accounts, now has the power,” the philanthropist said.

“Historically, countries had cyberattack teams and they could think about things like bioterrorism weapons,” he said. “Now small groups with AI can do what only larger countries can do. So it’s really bad intentions and AI that are emerging as a major threat. Unless governments insist on various safeguards, you know, we’ll see a lot of big, negative things because we’re not protecting.”

Gates specifically said that monitoring, “where you observe what the AI ​​is doing,” is an important safeguard to introduce into every AI lab.

“Law enforcement is going to be called for all of these fraud events or, you know, all of these transmission shutdown events,” he added. “And their ability to see what’s going on will be zero unless monitoring and safeguards are built into AI.”

Additionally, Gates said, these types of security features can be added to current AI systems without slowing the pace of development.

“It’s another debate: should we slow down or not,” he said. “But deciding whether to put safeguards and monitoring in place is a choice we have. And the only question is: do we do it before or after large-scale cyberattacks or bioattacks?”

A bipartisan Congressional proposal would require a “kill switch,” or a way for AI companies to immediately suspend or shut down their AI services if a dangerous situation arises.

Gates said “it’s not enough to have a circuit breaker” and renewed calls for internal oversight and safeguards.

“I would never want to say that I am against a circuit breaker,” he added. “But when you’re trying to moderate bad behavior, you need information and records of what’s being done. The kill switch alone won’t prevent these tragedies.”

The Gates Foundation and AI

The philanthropist, who now heads the Gates Foundation, also spoke about the group’s work to expand access to AI to countries where English is not the primary language.

“AI models have become very good at recognizing English and a dozen other common languages. So when you talk about your health, they’re going to record that accurately and be able to help you. In African languages, the error rate will be about 10 times higher,” he said, adding that mistakes can be “very dangerous.”

The Gates Foundation, he added, is working with 60 AI companies to expand the trusted languages ​​they offer.

“We have a very concrete timeline to improve quality, evaluate and then say, ‘OK, this is good enough. You can do a health interview or an educational session in that language,’” he added, hoping that within five years, every language spoken by a million or more people will be close to the quality of English on these platforms.

Gates said his foundation also uses AI in its own work.

“It helps me with foundation work,” he said. “You know, ‘What’s the latest news on malaria? What’s the latest news on malnutrition?’ This allows the Gates Foundation, which will spend all of its money over the next 19 years – you know, they have a little over $200 billion – that money is much better spent because we’re using AI to guide our work.

An “unfortunate” relationship with Jeffrey Epstein

Last year, Gates also came under scrutiny for his ties to the late Jeffrey Epstein, a disgraced financier and convicted sex offender. When the Justice Department released a slew of Epstein-related records at the direction of federal lawmakers, Gates’ name appeared several times, including in emails from Epstein suggesting Gates had extramarital affairs.

The Microsoft co-founder has since called his past relationship with Epstein “stupid” and testified before House lawmakers in June that he did not engage in illegal activities with Epstein.

On Wednesday, Gates accused Epstein of using their friendship to boost his reputation.

“Jeffrey told me after meeting him that he could help raise billions of dollars for global health – you know, he wouldn’t get paid, he wouldn’t have any role. It turned out to be a complete dead end. So spending time with him, like a lot of other people, kind of helped his reputation. You know, it’s very unfortunate,” he said.