Geoffrey Hinton Tells Congress It Has 'Maybe a Year' to Regulate AI Before Control Slips Away

Nobel laureate Geoffrey Hinton warned US senators in a closed-door briefing that Congress has roughly one year to establish AI safeguards, citing recursive self-improvement as the point of no return.

Geoffrey Hinton, the Nobel laureate widely known as the "Godfather of AI," delivered one of his starkest warnings yet to US lawmakers this week: Congress has roughly a year to put meaningful safeguards on artificial intelligence before the technology moves beyond human control. The British-Canadian computer scientist made the remarks on September 16 at a closed-door Senate briefing on Capitol Hill, according to NBC News.

"Maybe a year, but not much more than a year," Hinton told the room of lawmakers, according to NBC's account of the meeting. The briefing was convened by Senator Bernie Sanders and also featured MIT physicist Max Tegmark, co-founder of the Future of Life Institute, and AI safety researcher Ajeya Cotra, according to multiple reports of the session.

The Core of the Warning: AI Building AI

Hinton's argument centered on a specific technical milestone rather than a distant hypothetical. "AI has now reached the point where AI is designing better AI," he said, according to reports of the briefing. "That's called recursive self-improvement... it is going to get out of control unless we do something."

He also told lawmakers that expert timelines for superintelligence have collapsed over the past decade. "If you look at predictions for when we'll get superintelligence, it used to be maybe 30 years, maybe 50 years," Hinton said, per NBC News. "Then it came down to maybe 10 years, maybe 20 years. Now people are saying, a lot of the researchers are saying only a few years."

Hinton, a professor emeritus at the University of Toronto who shared the 2024 Nobel Prize in Physics for foundational work on neural networks, left Google in 2023 and has since become the field's most prominent internal critic — warning that the systems he helped create could end up outthinking and displacing their creators.

Skepticism of a 'Kill Switch'

Even as he urged rapid regulation, Hinton expressed doubt about one of the most-discussed proposals: a federal mandate that would let the government shut down dangerous AI systems. He told lawmakers such a "kill switch" would not hold up in the long run, according to CNN.

His skepticism landed the same week Senator Rand Paul blocked a Republican-led AI kill switch bill from being fast-tracked in the Senate — an episode that illustrates how far even minimal guardrail legislation remains from passage. Senator John Kennedy of Louisiana, who attended part of the briefing, told The Washington Sun he enjoyed the discussion but saw no sign the Senate would act, adding that he expects the next move to come from the White House, pointing to reporting that Treasury Secretary Scott Bessent is pushing for a safety framework.

The outlook across the aisle was not much sunnier. Senator Rick Scott of Florida said in a NewsNation interview that companies need to regulate themselves: "We're not going to get smart enough to regulate them. We're not going to get something done up here," he said, as reported by The Washington Sun.

Some Democrats pushed the other direction. Representative Yassamin Ansari of Arizona, who attended the briefing, wrote on X that AI "can and will continue to go beyond human control" and that "our window to regulate it is disappearing rapidly. We must pause NOW."

A Week of AI Safety Whiplash

The briefing capped a tumultuous stretch for AI governance debate. Days earlier, Anthropic CEO Dario Amodei published an essay calling on frontier labs to pace development, warning of risks from loss of control to bioterrorism — a position publicly backed by other lab leaders, according to widespread reports. The same day Hinton addressed senators, prominent AI researchers including Andrew Ng publicly dismissed extinction concerns as overblown, underscoring a field that is arguing with itself in public.

On Capitol Hill, the activity has been noisier than the output. Senator Cory Booker has called on President Trump to convene a special session of Congress on artificial intelligence, while a bipartisan Problem Solvers Caucus announced a new working group on AI and emerging technologies this week. None of it has yet produced binding federal law.

Why the One-Year Framing Matters

Deadline-driven warnings from AI pioneers have become routine, but the specificity of Hinton's timeline — delivered directly to the people with authority to legislate, in a room excluded to industry executives — marks a shift in tone. Previous warnings from researchers have typically projected risk on horizons of decades; a one-year window implies policy responses designed for the current Congress, not future ones.

It also sharpens the political problem. As Hinton's own kill-switch skepticism and Senator Paul's procedural block show, there is no consensus even among safety-minded figures on which interventions would work. Hinton has argued that the international situation — particularly competition with China — makes unilateral slowdowns politically fragile, a tension that has stymied most serious proposals so far.

What happens next is, by Hinton's own clock, a matter of months. Whether the briefing changes that calculus depends on a Senate that, by the accounts of its own members, is not currently moving at all.