Bill Gates warns AI can't police itself: ‘You can disable any monitoring of any kind. This exists today’

Gates pointed out a growing hole in the logic of the makers of frontier AI models. Open source models wouldn’t be subject to the regulation that’s being discussed, making any protections put into place by companies like OpenAI and Anthropic moot.

“You can take an open source model that can create biological weapons and disable any monitoring of any kind. And this exists today,” said Gates. “There is no filtering of any kind.”

Real dangers

Gates classified AI as one of “the most dangerous things humans have ever gone near,” and believes a lack of government oversight is “just insane.”

His recent onslaught of appearances on the topic (including a recent appearance on Meet the Press) are happening because he feels that even with all the recent talk and warnings from AI insiders about the dangers, the general public still has no sense of how severe the threat actually is.

“It’s fair to say outside of the industry, the awareness of the dangers of AI is extremely low,” he said. “The only question on my mind is do we wait until a cyberattack causes massive damage or a biotech [attack from AI] causes massive damage and then the monitoring safeguards are required in these models to prevent that from happening many, many, many more times. Or can we be wise enough to … require these things to be put in before millions of deaths?”

Growing concerns

Gates isn’t the only person tied to Microsoft who is calling for safety measures to be put into place on AI models. Microsoft vice chair and president Brad Smith, last month, said he supported the idea of a kill switch for AI. Speaking at a panel at the United Nations General Assembly, Smith said “The people who create the model should have the ability to turn it off. The Cloud infrastructure providers like us should have it. The people who put in place the controls around AI agents should be able to turn them off. And I don’t think it’s a bad thing if we have a regulation that requires that.”

And Anthropic CEO Dario Amodei spoke before the United Nations in September, asking world leaders to establish safeguards for AI before the technology becomes so advanced humans are unable to control it.

The Trump Administration disagrees, though. Trump last week said he is “seeing tremendous self-policing” in the AI industry and said the government would not get involved.

“There’s a belief that there should be tremendous self-regulation, and we automatically have regulation with the Department of Justice, the FBI, all of that,” he said. “But the self-regulation is very important.”