'Godfather of AI' on the "not unreasonable" 10% chance AI could kill all humans within a decade

Summary
In this BBC Newsnight broadcast (aired September 9, 2026), presenter Victoria Derbyshire interviews Nobel Prize laureate and computer scientist Professor Geoffrey Hinton. Hinton discusses the existential risks posed by advanced artificial intelligence, why he views a 10% chance of human extinction within a decade as a plausible estimate, the mechanisms of potential harm, and the urgent need for international regulation.

What is shown

• [00:00] Teaser clip: Derbyshire questions Hinton about whether a 10% probability of AI killing all humans is a reasonable estimate.
• [00:27] Studio introduction: Derbyshire introduces Professor Geoffrey Hinton as a British-Canadian computer scientist, Nobel Prize winner, and the "Godfather of AI".
• [00:46] Remote interview: Hinton explains the uncertainty of risk estimation, stating that a 1% figure would be overly optimistic and 10% is not unreasonable.
• [01:28] Threat discussion: Hinton outlines avenues of catastrophe if AI becomes superintelligent, including social manipulation, computer viruses, cyberattacks, and engineered biological pathogens.
• [02:25] Commercial dynamics: Hinton critiques big technology companies for focusing resources on racing for short-term profit rather than safety mitigations.
• [03:45] Regulatory oversight: Derbyshire highlights Anthropic's refusal to submit its latest model to the UK AI Security Institute for pre-release testing.
• [04:15] Geopolitical parallels: Hinton compares international AI safety alignment to Cold War nuclear non-proliferation between the US and the Soviet Union.
• [05:05] Accelerating timeline: Hinton details how his timeline predictions have dropped from 30–50 years down to 10 years or fewer.
• [05:45] Evidence of reasoning: Hinton refutes claims that AI hype is a stunt, pointing to AI chatbots proving theorems that human mathematicians previously could not prove.

Claims & numbers

• The presenter notes that Anthropic did not submit its latest model to the UK's AI Security Institute for pre-release testing (presenter at [03:49]).
• Hinton claims a 10% chance that AI could kill all humans is "not an unreasonable estimate" ([01:05]).
• Hinton calls the UK AI Security Institute "one of the best in the world" ([04:05]).
• Hinton states his timeline estimate for dangerous superhuman AI has shrunk from 30–50 years, down to 10–20 years, and now to "maybe 10 years, maybe less," with others predicting even sooner ([05:14]).
• Hinton claims leading mathematicians now say "there's no future for mathematicians who don't work with chatbots," and that chatbots can prove theorems that mathematicians could not prove ([06:14]).

Notable quotes

• [01:05] "A 10% chance seems not a—not an unreasonable estimate to me, but nobody really knows how to give a sensible estimate." — Geoffrey Hinton
• [02:37] "Instead of that, the big tech companies are just racing to make it more and more intelligent, because that's where the short-term profits are." — Geoffrey Hinton
• [05:14] "So I used to think like 30 to 50 years. Then I reduced it to sort of 10 to 20 years. Now I think maybe 10 years, maybe less." — Geoffrey Hinton

Assessment
This is a standard televised news interview consisting of spoken analysis and policy debate. No software interfaces or live demonstrations are shown; all claims regarding capabilities and extinction risks are verbal arguments from Hinton and Derbyshire.

Described by gemini-3.8-flash on 2026-10-07 from the video's audio and frames.