Story
September 26, 2026
Bill Gates Says AI’s Deadliest Risk Demands Government Action
Gates’s warning recasts AI not merely as a disruptive technology but as a security threat requiring public guardrails. The debate now turns on whether governments should compel those protections or leave companies to police themselves.
Bill Gates, once among the technology sector’s most prominent AI optimists, has moved sharply toward the safety camp. In comments previewing his appearance on NBC’s Meet the Press, the Microsoft cofounder said AI was “certainly powerful enough to drive events that cause a billion deaths.”1
His immediate concern is not an autonomous machine suddenly turning hostile, but people with hostile intent gaining access to increasingly capable tools. Gates called the combination of bad actors and frontier AI unprecedented in its destructive potential, and argued that political leaders must act before the systems become harder to contain.2
The prescription he outlined is regulation, not a blanket halt. “You need law enforcement and the politicians to get into the discussion about what safeguards and monitoring look like,” Gates said, describing those measures as a required cost rather than a dramatic brake on the industry.3
That position places him alongside a growing group of AI-safety advocates. Elon Musk amplified a post recalling his long-running warnings that AI presents an existential risk, including his past comparison of the technology to “summoning the demon.”
4 Anthropic chief executive Dario Amodei has likewise argued that regulation is the most effective way to pace frontier development, after reports that actors had used AI tools in planning drone attacks and biological research.1
But the industry and Washington remain divided over enforcement. Microsoft chief executive Satya Nadella has suggested companies should stop work themselves when they encounter a serious failure, while President Donald Trump has signalled little appetite for broad regulation. Gates’s intervention challenges that self-policing model: voluntary caution, he argues, is too thin a shield against a tool capable of amplifying catastrophic intent.
In an August essay, Gates said the world lacked a plan for a safe AI rollout and that he would support a credible global plan to slow advances. The latest warning gives that abstract concern a stark, human scale.1