Anthropic Researcher Resigns, Warning That AI Could Kill Us All
An Anthropic researcher has resigned over reckless AI development, while an internal alignment lead admitted to a greater than 10% chance of human extinction.
A fresh wave of alarm has hit the artificial intelligence industry following the departure of a researcher who accused leading firms of recklessness. On Tuesday, September 9, 2026, Anthropic researcher Jacob Coxon announced his resignation, warning that the race to develop increasingly powerful AI systems could pose an existential threat to humanity. Coxon spent three years working on AI pretraining research at both OpenAI and Anthropic.
In a series of posts on X, Coxon stated that neither company is acting responsibly, writing that developers are racing straight to self-improving superintelligence and gambling with human lives. He asserted that "the people building AI earnestly believe that it could kill us all by the end of the decade," and noted that this is not a marketing stunt.
Evan Hubinger Responds to Resignation
Coxon’s statements received immediate reaction from within Anthropic. Evan Hubinger, Anthropic's alignment science lead, stated on X that Coxon was correct, writing: "Jacob is correct here; we really do earnestly believe all humans! I personally think it is >10 per cent within the next decade." Hubinger clarified that this estimate represents his personal view rather than an official company position. Hubinger stressed that current models present low risk, but expressed deep concern over future superintelligence arising from recursive self-improvement.
| Detail | Information |
|---|---|
| Resigning Researcher | Jacob Coxon |
| Companies Where Coxon Worked | OpenAI and Anthropic |
| Evan Hubinger's Risk Estimate | Greater than 10% chance of human extinction within the next decade |
| Dario Amodei's Prior Risk Estimate | 25% odds of AI going "really, really badly" |
Other industry leaders have previously voiced similar concerns. Anthropic CEO and co-founder Dario Amodei signed a statement in 2023 declaring that mitigating the risk of extinction from AI should be a global priority. Amodei has also stated publicly that he puts the odds of AI going "really, really badly" at 25%.
Regulatory Context and Industry Scrutiny
The public admissions from internal researchers have intensified policy debates. U.S. Representative Ted Lieu pointed to the developments to argue for hard legal guardrails on AI, while financial and political figures weighed in on the turbulence surrounding the rapid transition to the AI era.
At the same time, Anthropic faces separate scrutiny regarding its engagement with oversight bodies. According to the Financial Times, Anthropic declined to give the U.K. AI Security Institute access to testing ahead of its release, as the tech giant appeared to cede to the Trump administration's demands to not make models available to foreign nationals.
Frequently Asked Questions
Who is Jacob Coxon?
Jacob Coxon is a pretraining researcher who spent three years working at both OpenAI and Anthropic before resigning on Tuesday, September 9, 2026, citing AI risks.
What did Evan Hubinger say about extinction risk?
Evan Hubinger, Anthropic's alignment science lead, stated that he personally believes there is a greater than 10% chance that advanced AI could kill all humans within the next decade.
As Anthropic and other labs continue their work, the debate over whether any single lab can safely build self-improving intelligence without government intervention or a coordinated slowdown remains unresolved.
CurrentDeck is dedicated to independent, evidence-backed reporting. This briefing was synthesized from primary source reporting, corroborated across independent newsrooms, and verified against our Editorial Standards.
