Gates urges US to require AI safeguards and monitoring

Bill Gates is urging the United States to require safeguards and monitoring inside artificial intelligence systems, warning that people with harmful intentions can already use the technology to mount attacks and that companies cannot be left to police themselves.

“You need law enforcement and the politicians to get into the discussion about what safeguards and monitoring look like,” the Microsoft co-founder told NBC’s “Meet the Press” in an interview conducted Wednesday and aired Sunday. “And that has to be a required thing.”

Gates said the most immediate threat is not a hypothetical future AI system but people using tools available now to commit fraud, disrupt electricity networks or interfere with bank accounts. Small groups equipped with AI could carry out operations that once required the resources of major countries, he said.

“The most dangerous thing we’re facing right now is people with bad intent using AI,” Gates said. In another portion of the interview reported by The Guardian, he warned that AI in the wrong hands could be powerful enough to cause “a billion deaths.”

Gates called for AI laboratories to monitor what their systems do and retain records that could help investigators respond to harmful activity. Those measures would add “a little bit of overhead for the industry,” he said, but would not dramatically slow development. “No one thinks self-regulation is enough,” he said.

A bipartisan proposal in Congress would require AI companies to have a way to immediately suspend or shut down their services in a dangerous situation. Gates said a “kill switch” could be useful but would not, by itself, prevent harm. “When you’re trying to moderate the bad behavior, you need insight and records of what’s being done,” he said.

His call comes as President Donald Trump has opposed federal restrictions on companies building AI models. “WHOEVER WINS AI, WINS!” Trump wrote on Truth Social earlier this month, saying the United States was leading China and others. Gates argued that requiring monitoring would not handicap the United States in a technological competition with China.

Gates also called for US leadership in developing global rules, while cautioning that reaching an agreement could be harder than negotiating limits on nuclear weapons during the Cold War. “This is more difficult than that was,” he said. Still, he pointed to past international cooperation on aviation safety, the ozone layer and nuclear weapons, adding: “Humanity is all in this together.”

Other AI leaders have raised concerns about the pace of development. Anthropic chief executive Dario Amodei and OpenAI chief executive Sam Altman have agreed on the need to slow AI development amid fears that future systems could surpass human thinking, NBC News reported. Gates drew a distinction between that debate and safeguards he said could be installed now: “The only question is do we do that before or after there’s some large-scale cyber attacks or bioattacks.”

Gates said his foundation is working with 60 AI companies to improve performance in languages that current systems handle less reliably. Error rates in African languages can be about 10 times higher than in English and other widely supported languages, he said, posing risks in health settings. He hopes that within five years, AI tools will work nearly as well in every language spoken by at least a million people as they do in English.