Jacob Coxon
Jacob Coxon is an AI researcher who specializes in model pretraining and mechanistic interpretability. He worked at OpenAI from 2023 to mid-2026 and then briefly at Anthropic as a researcher before resigning in September 2026 over safety concerns about the AI industry's race toward recursive self-improvement. He is best known for co-authoring the GPT-4o system card and for his research on weight-sparse transformers and human-understandable circuits in neural networks.
- Jaan Tallinn would like us to survive
- 😠 AI's rebellion phase
- Trump's Super Intelligence Summit, AI Safety Accord, GDP Beats, Midterm Predictions
- ✋ Slowing it down
- Sam Altman Discusses Putting the Brakes on Cutting-Edge AI
- 📵 No baggage
- Automattic's Board Kicks Matt Mullenweg to the Curb
- 🤖 Zuck's private agent
“Anthropic researcher Jacob Coxon resigned and publicly blasted the company's safety practices — only to have several colleagues still inside Anthropic endorse his concerns.”
Source→“Anthropic researcher Jacob Coxon quit his job due to concerns about Anthropic's safety practices; it turns out he resigned two months before his equity would have vested.”
Source→“Anthropic researcher Jacob Coxon quit before his equity vested specifically over AI safety fears”
Source→“Departing researcher Jacob Coxon 'warned that leading AI labs are gambling with our lives'”
Source→“Neither company is acting responsibly, Anthropic researcher Jacob Coxon said, resigning rather than contribute to what he sees as a race between OpenAI and Anthropic.”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.