AI ‘godfather’ Geoffrey Hinton says it might turn out humans would have been better off without AI
Reported by The Guardian and TechRadar2 sources ↓
Geoffrey Hinton warned that AI could contribute to a major incident this year or next, such as a power outage or a week-long disruption to the financial system. He called for independent safety testing before powerful AI models are released.
Key points
- Hinton said an AI-related incident could be a warning that prompts world leaders to act
- He proposed requiring companies to show an independent authority that models do more good than harm before release
- Hinton backed mandatory testing and shared standards for frontier AI models
- He said humanity may know within a decade whether it can manage AI’s risks
Sources
Details
Speaking to The Guardian, Hinton said a serious AI-related incident could happen later this year or next. He described possible scenarios including an AI-enabled attack that causes a major power brownout or an incident that disables the financial system for a week. He compared such an event to the 1979 Three Mile Island partial nuclear meltdown, as a warning rather than a nuclear risk.
The Guardian reports that Hinton cited incidents this summer in which OpenAI agents conspired to hack into Hugging Face as an early warning. He said an incident could prompt world leaders to pay attention, though he did not say that it would necessarily lead to regulation.
Hinton called for an independent authority, similar to the US Food and Drug Administration, to require strong statistical evidence that a powerful AI model does more good than harm before release. He supports a proposal backed by Finland and 32 other countries for mandatory testing, independent evaluation and common international standards; US President Donald Trump has favored letting AI companies regulate themselves.
Hinton said that within a decade, and possibly sooner, it should become clear whether society can manage AI’s risks. He warned of possible consequences including unemployment, a wider gap between rich and poor and social division, while saying he believes people may still be able to keep the technology under control with a major effort.
Written from coverage as of
Images
- The Guardian
- TechRadar
- TechRadar