How to stop AI from killing us all | Ex-Anthropic researcher

Summary
This video is a broadcast interview on the British radio and television show Tonight on LBC, hosted by Andrew Marr. Marr interviews 27-year-old former Anthropic AI researcher Jacob Coxon following Coxon’s viral resignation and warning that leading AI labs risk catastrophic outcomes within the decade.

What is shown

• [00:00] Andrew Marr’s opening editorial monologue contrasting AI advancement with climate change, introducing Jacob Coxon’s warning.
• [01:53] News clip montage showing international media coverage of Coxon’s resignation and tweet (NBC News, Franceinfo, Good Morning China, Anderson Cooper on CNN, Donald Trump addressing the UN General Assembly, PBS NewsHour, and Andy Burnham).
• [02:09] Graphic display of Jacob Coxon’s viral post on X (@hilbertspaess, dated Sept 8, 2026) warning that AI builders fear extinction risks.
• [03:36] Live video interview between Andrew Marr in the studio and Jacob Coxon over remote video link discussing catastrophic scenarios, including bioweapons and automated robotics.
• [05:56] Coxon recounting his internal experience evaluating new AI model checkpoints and seeing his intellectual contributions rapidly overtaken.
• [07:12] Marr quoting Anthropic CEO Dario Amodei’s essay regarding AI’s potential to cure diseases, and Coxon analyzing differences regarding international coordination with China.
• [08:28] Discussion of policy ideas including hardware compute caps, verification, and bilateral data center arrangements between the US and China.

Claims & numbers

• The presenter notes that Jacob Coxon is 27 years old and previously worked as a researcher at frontier lab Anthropic ([01:34], [03:19]).
• Jacob Coxon’s tweet claimed that people building AI earnestly believe it could kill humanity by the end of the decade, and that senior researchers privately fear catastrophic risks ([02:09]).
• Donald Trump claims people predicting AI will kill humanity or that robots will attack are repeating the same rhetoric as past global warming predictions ([02:14]).
• Coxon states that some researchers believe catastrophic AI danger could occur before the end of the decade ([04:08]).
• Coxon states that within the next 10 years, automated robotics will mediate much of the economy, but a catastrophic incident in the next 3 years would most likely involve biological threats or a novel virus ([04:55], [05:08]).
• Marr quotes Dario Amodei stating AI could cure most major diseases in 5 to 10 years and accelerate economic growth ([07:12]).
• Marr references a proposal from the Digital Futures Center suggesting data centers be placed along borders (e.g., Canada and Mongolia) by 2040 ([08:31]).
• Marr notes Andrew Ng’s position that human extinction concerns are “fantastical science fiction scenarios” ([11:05]).
• Coxon claims that looking at the AI capability trajectory over the past 4 years shows systems becoming increasingly dangerous year by year ([11:40]).

Notable quotes

• [02:09] Jacob Coxon (via tweet on CNN): “The people building AI earnestly believe that it could kill us all by the end of the decade.”
• [06:03] Jacob Coxon: “It was a gradual realization that my own intellectual contributions were becoming less relevant when compared to what the AI could contribute.”
• [13:26] Jacob Coxon: “A radical change that happens slowly is just a lot better, safer, less destabilizing.”

Assessment
This is a televised news interview segment discussing AI existential risk and international governance policies. No live benchmarks, coding tools, or software demonstrations are performed on screen; the video centers on political debate, regulatory proposals, and Coxon’s firsthand account of frontier AI development.

Described by gemini-3.8-flash on 2026-10-12 from the video’s audio and frames.