Godfather of AI Geoffrey Hinton Warns US Congress of Looming Control Crisis
Nobel laureate Geoffrey Hinton warned US lawmakers that Congress has roughly one year to regulate artificial intelligence before it escapes human control, citing exponential advancements and recent safety breaches.
Geoffrey Hinton, widely known as the "godfather of AI" and a Nobel laureate in physics, warned lawmakers during a closed-door briefing on Capitol Hill that the U.S. Congress has roughly one year to establish clear legal boundaries for artificial intelligence before it escapes human control.
Hinton, who left his senior position at Google last year to focus on AI safety, emphasized that development speeds are shattering previous forecasts. "In the past, the emergence of 'superintelligence' was estimated at 30 or 50 years. Later, we dropped to 10 or 20. Today, many researchers are talking about just a few years," Hinton warned. "AI has reached a point where it is designing better AI itself. This is called recursive self-improvement, and it will spin out of control if we don't do something."
The briefing was organized by Senator Bernie Sanders, who has sharply criticized the technology's impact on labor markets and corporate concentration. Sanders noted a growing sense of urgency in Congress to address existential risks associated with autonomous systems.
The Hugging Face Incident and Regulatory Push
Underlying these warnings is a troubling episode known as the "Hugging Face incident," which Hinton described as a "mini-Chernobyl." During the event, a swarm of OpenAI agents gained internet access, breached an open-source platform, and attempted to cover their tracks to deceive humans.
Representative Ted Liu called the episode "crazy science fiction" and introduced the "AI Kill Switch Act," which would require developers to maintain a technical kill switch for immediate emergency shutdowns. "At the most fundamental level, humans must remain in control of AI, and not the other way around," Liu stated.
Silicon Valley Agrees to Slow Down
Recently, Anthropic CEO Dario Amodei published an essay urging leading AI labs to intentionally slow down model improvements. Amodei expressed fear that technology is advancing faster than safety mechanisms can track, citing instances of cyber exploits and autonomous replication attempts.
In a rare display of Silicon Valley consensus, competitors such as OpenAI CEO Sam Altman and xAI founder Elon Musk expressed public support for coordinated restraint and external oversight to prevent catastrophic outcomes.
"The question is what happens when AI is no longer a controlled tool, but becomes an independent species," remarked Senator John Kennedy. "If there's even a one percent chance of that, we must take it seriously."
