Anthropic Insiders Warn AI Could End Humanity This Decade
Three Anthropic researchers warn AI may cause human extinction this decade.
Why it matters: This rare insider alarm spotlights growing ethical tensions in AI development and governance. Legal professionals must track emerging risks related to AI liability and regulatory oversight.
- Jacob Coxon resigned from Anthropic on Sept. 8, 2026, citing AI risks.
- Evan Hubinger estimates over 10% chance AI could kill all humans within a decade.
- Anthropic is racing to develop self-improving superintelligent AI, raising safety concerns.
- Protect Democracy sued four US agencies over secretive Trump-era AI safety framework.
Three researchers from AI startup Anthropic have publicly warned of the grave dangers posed by artificial intelligence, predicting that current advancements could lead to human extinction within the next decade. Jacob Coxon, who resigned on September 8, 2026, cited concerns that AI development is a ‘‘hubristic gamble’’ with potentially catastrophic consequences, stating, “The people building AI earnestly believe that it could kill us all by the end of the decade.”
Evan Hubinger, Anthropic’s Alignment Science Lead, put a quantifiable estimate on that risk, warning of a >10% chance that AI could wipe out humanity within 10 years (Forbes).
This insider alarm comes amidst a highly competitive race among AI companies to develop self-improving superintelligent systems, which has raised widespread concerns about safety, oversight, and ethical considerations (Washington Post).
Meanwhile, the Trump administration’s AI framework has faced criticism for its lack of public incident reporting guidelines following incidents like the Hugging Face breach. Protect Democracy has taken legal action against four US federal agencies, alleging the administration’s secretive approach compromises transparency and public accountability (Axios, TechRadar).
These developments underscore urgent legal and regulatory challenges ahead. AI governance frameworks will need to address risks of existential harm and enforce transparency in an industry pushing the boundaries of technology at unprecedented speed.
By the numbers:
- September 8, 2026 — Jacob Coxon resigns from Anthropic
- 10% — Evan Hubinger's estimated chance AI could kill all humans within the decade
- 4 — Number of federal agencies sued over AI safety framework secrecy
What's next: Watch for regulatory responses to lawsuits on AI framework transparency and further insider disclosures about AI safety.