Gates warns Trump inaction leaves world exposed to AI pandemic risk
What's the deal? Bill Gates has accused the Trump administration of blocking efforts to guard against AI's risks, warning the technology could be used to cause a pandemic "significantly worse than Covid." The Microsoft co-founder spoke on Wednesday at the Financial TimesDealroom has a profile for this one. Try Dealroom →' Future of AI Summit.
What's the concern? Gates said bad actors using AI to build bioweapons or exploit cyber vulnerabilities pose the greatest immediate risk. He flagged the danger of so-called "open" models that are freely accessible to developers, whose safeguards and monitoring can be stripped out.
An estimated 15 million deaths were linked to Covid in 2020 and 2021; Gates has previously warned AI could cause a billion deaths.
Why the US? International efforts to regulate and monitor AI are being hindered by White House inaction, Gates said. President Donald Trump has rebuffed calls for stricter rules, dismissing an attempt to develop G20 AI guidelines as a "globalist scheme" and favouring self-regulation.
"There's not much value in acting unless the US is part of that. So we're kind of stymied," Gates said. "The only question is how many bad events need to take place before they change their posture."
What's he asking for? Gates called for a benchmark for models' internal safeguards, third-party auditing of new models, and a US-led global effort to fix cyber vulnerabilities. He also urged "massive" monitoring of "air, sewage, blood" to catch AI-enabled infections early.
Self-regulation would not work, he said: "As soon as we add multiple companies, then the only way to solve this moves up to the government."
Why now? The warning lands as concern over AI safety intensifies. Anthropic's IPO prospectus last week flagged potential "catastrophic or existential risks to humanity," which Gates praised: "They've always been very open, and this is another example of that."
He expects OpenAI and Anthropic to release new models this year, plus "two major releases" in 2027. The risk of "the models themselves being out of control" could become serious with those upcoming releases, he added.
The signal: Gates is pushing for binding government standards rather than voluntary pledges — the gap at the centre of current US policy, which rests on industry commitments without clear enforcement. His intervention is a risk warning, not evidence of a specific AI-enabled bioweapon, but it sharpens a core governance tension: protections built into a provider's service may not survive once models are freely accessible and modifiable.
Read more: Financial Times
Image credit: Generated with Gemini