I resigned from Anthropic today
- ID
- 22655
- Status
- summarized
- Published
- 09 Sep 2026, 8:40 AM
- Fetched
- 11 Sep 2026, 5:45 AM
- Provider
- Hacker News
- Category
- dev-community
- Original URL
- https://twitter.com/hilbertspaess/status/2097476196791709843
- Source URL
- https://hnrss.org/best
Summary
- Score
- 4.0
- Created
- 11 Sep 2026, 5:47 AM
- Tags
- Audience
- ai_ml_learnerssaas_founders
What happened
Jacob Coxon announced his resignation from Anthropic, claiming he spent three years doing pretraining research at both OpenAI and Anthropic and that neither company is acting responsibly. He alleges both labs are racing toward self-improving superintelligence, that senior researchers privately express fear of civilizational risk, and that Anthropic is locked in a race dynamic despite understanding the stakes. He references a 'Hugging Face attack' as a warning shot that has made pacing agreements more viable.
Why it matters
This is an insider's advocacy claim, not a technical disclosure or actionable finding. There is no concrete evidence, incident detail, or specific technical risk described that would change what a builder should do today. The vague reference to a 'Hugging Face attack' is unexplained and cannot be acted on. Treat this as a signal of internal lab culture tension and AI safety debate, not as guidance for tooling, infrastructure, or product decisions.
Discussion angle
Does this resignation thread change your risk calculus when building on Anthropic or OpenAI APIs, or is it advocacy that doesn't affect day-to-day engineering and product decisions?