> ## Content Index
> Fetch the complete content index at: https://globalfeed.ai/llms.txt
> Use this file to discover other available public pages before exploring further.

# Anthropic safety lead puts odds of AI killing all humans above 1 in 10
- URL: https://globalfeed.ai/en/anthropic-safety-lead-puts-odds-of-ai-killing-all-humans-above-1-in-10/
- Published: 2026-09-09T14:30:44.000Z
- Updated: 2026-09-09T14:30:44.000Z
- Description: Evan Hubinger, who leads an Anthropic safety team, says the chance AI kills all humans within the decade is greater than 10 percent, hours after a colleague resigned over safety.
- Author: GlobalFeed Editor
- Tags: Anthropic, OpenAI, AI safety, x-AnthropicAI, dil-en, denetim-gerekli, video

Evan Hubinger, who leads one of Anthropic's AI safety teams, said he personally estimates a greater than one in 10 chance that AI "could kill all humans" within the next decade. He spoke hours after a colleague resigned over safety concerns at the lab.

Jacob Coxon, a researcher who trained AI systems at Anthropic and previously did so at OpenAI, announced his departure in a post on X, saying he quit over the company's lax approach to safety. Coxon accused both firms of "racing straight to self-improving superintelligence and gambling with our lives," even though "the people building AI earnestly believe that it could kill us all by the end of the decade."

## 'Not yet have a plan'

Responding directly, Hubinger said he worries about self-improving AI and that it "is happening faster than we thought." He agreed with Coxon's characterisation: "We really do earnestly believe AI could kill all humans."

Even so, Hubinger said Anthropic does "not yet have a plan" for keeping advanced AI safe and aligned with human values, and is "not clearly on track to" produce one. Coxon said the companies are "locked in a race" to build advanced systems first and are pushing ahead "despite the risk."

Industry figures have warned for years that self-improving systems could spiral beyond human control in a loop known as recursive self-improvement. That has not happened, but companies are actively pursuing the goal, and much of today's AI code is written with help from AI.

Coxon's exit is one of the most visible departures from Anthropic, which was founded by former OpenAI staff after safety disagreements there. Several researchers have cited safety concerns when leaving OpenAI in recent years.

Full report at [The Verge](https://www.theverge.com/ai-artificial-intelligence/991927/anthropic-ai-kill-all-humans?ref=globalfeed.ai).