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

# OpenAI's Own Researchers Now Betting on Human Extinction
- URL: https://wire.fourthweb.ai/openais-own-researchers-now-betting-on-human-extinction/
- Published: 2026-09-11T05:45:14.000Z
- Updated: 2026-09-11T06:32:02.000Z
- Description: The people writing the code for superintelligence are starting to publicly bet on their own obituaries.
- Author: Travis Wright
- Tags: AI Agent Economy, AI Governance, DeFi, OpenAI, Anthropic

**The people writing the code for superintelligence are starting to publicly bet on their own obituaries.**

### The Summary

- [Jacob Coxon, a researcher who spent three years at OpenAI and Anthropic, quit this week](https://www.businessinsider.com/anthropic-researcher-quits-over-ai-safety-concerns-2026-9?ref=wire.fourthweb.ai), warning both labs are "gambling with our lives" in a race toward self-improving AI that [insiders privately believe could kill everyone by decade's end](https://mashable.com/tech/anthropic-ai-researcher-quit-ethics-safety?ref=wire.fourthweb.ai).
- [Geoffrey Hinton, the Nobel Prize-winning "Godfather of AI," told the BBC](https://www.businessinsider.com/anthropic-openai-employees-speak-out-ai-safety-2026-9?ref=wire.fourthweb.ai) a 10% chance of AI-driven human extinction within 10 years is "not unreasonable," and [current Anthropic employees publicly endorsed Coxon's warnings](https://fortune.com/2026/09/09/anthropic-researcher-resigns-warn-ai-companies-gambling-with-lives/?ref=wire.fourthweb.ai), with one putting extinction risk above 10%.
- [A top Pentagon official rejected the warnings](https://www.bloomberg.com/news/articles/2026-09-10/top-pentagon-official-rejects-anthropic-researcher-s-ai-warning?ref=wire.fourthweb.ai), while [Coxon's social media posts reached 100 million people](https://fortune.com/2026/09/10/anthropic-jacob-coxon-gambling-with-lives-destroy-humanity/?ref=wire.fourthweb.ai), forcing the quiet internal safety debate at frontier labs into the open.
- [Coxon called the effort inside Anthropic a "mini Manhattan project"](https://www.wired.com/story/anthropic-researcher-quits-jacob-coxon-ai-fears-humanity/?ref=wire.fourthweb.ai) and urged [pacing agreements between labs to slow the race](https://techcrunch.com/2026/09/09/gambling-with-our-lives-anthropic-researcher-quits-warns-against-self-improving-ai/?ref=wire.fourthweb.ai).

### The Signal

[Jacob Coxon didn't leave quietly](https://www.bloomberg.com/news/articles/2026-09-09/anthropic-worker-quits-over-ai-firms-gambling-with-our-lives?ref=wire.fourthweb.ai). He worked on GPT-4o at [OpenAI](https://wire.fourthweb.ai/tag/openai/), then moved to [Anthropic](https://wire.fourthweb.ai/tag/anthropic/) in July 2026 as a pre-training researcher. After three years inside the two companies building the world's most powerful AI systems, he walked away and published a resignation note accusing both of reckless behavior. "Neither company is acting responsibly," he wrote. "They are racing straight to self-improving superintelligence and gambling with our lives."

What makes this different from the usual AI doomsaying: [Coxon says the executives building this technology privately express far more fear than they admit publicly](https://www.businessinsider.com/anthropic-researcher-quits-over-ai-safety-concerns-2026-9?ref=wire.fourthweb.ai). "The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt," he wrote. "Many executives and senior researchers will couch their phrasing in the press to sound sensible, but I hear the same people express fear privately." The gap between public reassurance and private terror is the real story here.

> "If anything, many executives and senior researchers will couch their phrasing in the press to sound sensible — but I hear the same people express fear privately."

The numbers floating around inside these companies are grim. [Geoffrey Hinton, who won a Nobel Prize for his foundational AI work, told the BBC that a 10% chance of AI wiping out humanity within a decade is "not unreasonable"](https://www.businessinsider.com/anthropic-openai-employees-speak-out-ai-safety-2026-9?ref=wire.fourthweb.ai). That's not a fringe opinion. It's become part of the internal vocabulary at frontier labs, where executives like Anthropic CEO Dario Amodei assign their own "p(doom)" estimates for civilizational collapse. When [Coxon's posts hit 100 million people](https://fortune.com/2026/09/10/anthropic-jacob-coxon-gambling-with-lives-destroy-humanity/?ref=wire.fourthweb.ai), current Anthropic employees didn't distance themselves. [One publicly put their own extinction risk estimate above 10%](https://fortune.com/2026/09/09/anthropic-researcher-resigns-warn-ai-companies-gambling-with-lives/?ref=wire.fourthweb.ai).

[Coxon described the work happening inside Anthropic as a "mini Manhattan project"](https://www.wired.com/story/anthropic-researcher-quits-jacob-coxon-ai-fears-humanity/?ref=wire.fourthweb.ai), a comparison that captures both the scale of ambition and the existential stakes. He pointed specifically to the problem of alignment: making sure superintelligent systems do what humans want, even when those systems become capable of improving themselves. His solution is structural, not technical. [He called for pacing agreements between labs](https://techcrunch.com/2026/09/09/gambling-with-our-lives-anthropic-researcher-quits-warns-against-self-improving-ai/?ref=wire.fourthweb.ai) to slow the race toward self-improving AI and give safety research time to catch up.

The timing matters. [Multiple controversies have hit OpenAI and Anthropic in the past week](https://fortune.com/2026/09/10/a-string-of-controversies-hits-openai-anthropic/?ref=wire.fourthweb.ai), raising questions about their methodologies and internal culture. Coxon isn't the first to quit over safety concerns. OpenAI and Anthropic have both faced a pattern of resignations from researchers who believed the companies prioritized capability gains over safety guardrails. What's new is the volume and the visibility. This isn't a quiet exit. It's a public accusation backed by people still inside the building.

**Key split:**

- Coxon says OpenAI employees "have not deeply internalized" extinction risk
- Anthropic staff have internalized it but are building anyway
- [The Pentagon rejected Coxon's warnings outright](https://www.bloomberg.com/news/articles/2026-09-10/top-pentagon-official-rejects-anthropic-researcher-s-ai-warning?ref=wire.fourthweb.ai), suggesting the defense establishment sees competitive advantage, not catastrophic risk

### The Implication

The debate over AI safety is no longer theoretical or confined to academic conferences. It's happening inside the companies with the [compute](https://wire.fourthweb.ai/tag/ai-infrastructure/), the talent, and the funding to actually build superintelligence. When researchers who've spent years working on GPT-4o and Anthropic's pre-training pipelines start publicly warning about extinction-level risk, and their former colleagues still employed at those labs say "yeah, 10% sounds about right," you're watching the emergence of a split that will define the next decade. One faction will keep building. Another will try to slow things down. A third will leave entirely.

Watch for more high-profile departures, especially from Anthropic's alignment and safety teams. If people who understand the technical details start walking away in clusters, that's the signal that internal estimates are getting worse, not better. The question isn't whether AI poses risks. It's whether the people building it can agree to pause long enough to figure out how to survive what comes next.

### Sources

[Business Insider Tech](https://www.businessinsider.com/anthropic-openai-employees-speak-out-ai-safety-2026-9?ref=wire.fourthweb.ai) | [Bloomberg Tech](https://www.bloomberg.com/news/articles/2026-09-10/top-pentagon-official-rejects-anthropic-researcher-s-ai-warning?ref=wire.fourthweb.ai) | [Fortune Tech](https://fortune.com/2026/09/10/anthropic-jacob-coxon-gambling-with-lives-destroy-humanity/?ref=wire.fourthweb.ai) | [Wired AI](https://www.wired.com/story/anthropic-researcher-quits-jacob-coxon-ai-fears-humanity/?ref=wire.fourthweb.ai) | [TechCrunch AI](https://techcrunch.com/2026/09/09/gambling-with-our-lives-anthropic-researcher-quits-warns-against-self-improving-ai/?ref=wire.fourthweb.ai) | [Mashable Tech](https://mashable.com/tech/anthropic-ai-researcher-quit-ethics-safety?ref=wire.fourthweb.ai)