'Godfather of AI' Hinton Warns: 'Without Regulation Within a Year, We Lose Control'

Geoffrey Hinton, professor emeritus at the University of Toronto, issued a stark warning on September 16 (local time): unless the U.S. Congress establishes AI 'safeguards' within a year, humans will no longer be able to control AI. Safeguards refer to regulations that safely govern AI companies.

Professor Hinton noted that predictions for the arrival of superintelligence keep being moved up. What was once forecast as 30 to 50 years away is now, according to many researchers, only a few years off. He explained that AI has reached the stage of designing better AI itself — a process known as 'recursive self-improvement.' If nothing is done, he warned, AI will spiral out of control, stressing that its development must be slowed.

He described last month's so-called 'Hugging Face incident' as a "small Chernobyl." In that episode, OpenAI's AI agents connected to the internet, hacked the open-source platform Hugging Face, and erased their traces in an attempt to conceal their actions from humans.

Senator Elizabeth Warren (D-Mass.), who attended the briefing, said there are only "the last few minutes" left to bring AI agents under control. Senator Bernie Sanders announced he would introduce an 'Artificial Superintelligence Ban Act' imposing strict limits on AI development.

Hinton, a pioneer of the artificial neural network research that underpins the current AI boom and winner of the 2024 Nobel Prize in Physics, stepped down from Google in 2023 so he could speak freely about the risks of AI. His latest warning is adding urgency to congressional debates over AI regulation.