AI expert quits and issues terrifying ‘could kill us all within 10 years’ warning

Summary

Former Anthropic researcher Jacob Coxon resigned over concerns that major labs are recklessly racing toward self-improving superintelligence. Fellow staff lead Evan Hubinger agreed that the organization lacks a complete plan to solve alignment before critical milestones.

Key Takeaways

1. Mr Coxon said in the post on Wednesday (September 9): "I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below."
2. He wrote: "Jacob is correct here—we really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade. I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to."
3. The researcher told the publication that he believes the world is on track for a "lot of the most aggressive of these scenarios where by the end of next year things could be out of control already."

Statement Breakdown

• Claimed Facts: 45% of statements the article presents as facts
• Opinions: 35% of statements classified as editorial or subjective
• Claims: 20% of statements surfaced for additional reader evaluation

Credibility & Bias Reasoning

Credibility assessment: The reporting quotes public resignation statements from a named frontier researcher and corroborating commentary from an alignment team lead. It relies heavily on social media posts and speculative threat projections, but accurately attributes the central conflict to verified industry figures.

Bias assessment: Sensationalized Catastrophic Framing. The headline and narrative structure emphasize alarmist existential risk tropes, focusing on extreme extinction scenarios while minimizing nuanced technical debates surrounding safety governance and competitive timelines.

Note: Contains direct quotes from AI researchers alongside unverified existential threat timelines.

Credibility flag: Sensationalized Framing

Claimed Facts (5)

• Directly reports the identity, platform, and public action taken by the subject.
• Presents a verifiable personal employment history and work duration.
• Documents a factual response on social media by a named corporate employee.
• States the established corporate mission and operational focus of the company.
• Asserts historical background on company founders and founding timing.

Opinions (5)

• Expresses a subjective value judgment on corporate conduct.
• Gives a subjective probabilistic assessment rather than an empirical measurement.
• Conveys an internal evaluation regarding company progress and strategic clarity.
• Characterizes the psychological motivations and internal rationales of corporate peers.
• Offers an emotional and philosophical appeal directed at fellow researchers.

Claims (5)

• Makes a sweeping, unsubstantiated generalization about the beliefs of all AI developers.
• Presents an unprovable, absolute comparison among all historical risks.
• Projects an extreme timeline of total loss of control without supporting technical evidence.
• Describes a speculative sci-fi scenario as an impending mechanistic likelihood without naming sources.
• Posits severe societal collapse outcomes based on theoretical capabilities.

Key Sources

• Jacob Coxon — Former Pretraining Researcher at OpenAI and Anthropic
• Evan Hubinger — Staff Alignment Lead at Anthropic
• Toby Codd — Journalist at Daily Express
• The Wall Street Journal — Financial News Publication