Jacob Coxon
Jacob Coxon is an AI researcher who worked on pretraining at OpenAI and Anthropic. He left Anthropic in 2026 while warning publicly about risks from the frontier AI race.
- Videos
- 7
- Channels
- 7
- Topics
- 9
0 comments · 0 votes
Video appearances
Video topic coverage
Based on 7 reviewed videos for Jacob Coxon; compared with all 2317 public reviewed videos. Catalogue coverage, not expertise or speaking time.
Unmapped topics: 0 for this person; 12 in all videos. Scores cover mapped topics only.
- Jacob Coxon (7 videos; normalised to 100%)
- All videos (2317; normalised to 90%)
Both plots connect neighbouring topic values in the same ten-axis order, including zeros at the centre. The blue person plot is proportionally normalised to a 100% maximum and the white baseline to 90%. Each plot uses its own scale, showing relative topic emphasis; the values below show actual coverage.
- Agents & workflows0%57.7%
- Coding & tools0%40.6%
- Models & learning38.6%38.5%
- Infrastructure & compute0%23.2%
- Research & evaluation45.7%40.6%
- Safety & security100%30.2%
- Business & productivity0%29.3%
- Creative & media0%7.3%
- Robotics & perception0%5.5%
- Society & policy100%17.8%
Topic attribution
Both datasets use the same explicit fractional topic memberships. For each video, only the strongest membership in each group counts. We average those values across distinct public reviewed videos, including untagged videos. The all-video baseline includes this person's videos. Both datasets come from the same active catalogue snapshot and are independent of filters and viewing history. Groups overlap, so percentages do not need to total 100. Weights describe topic membership, not measured speaking time or relationship confidence.
Topics associated with this person:
- AI Alignment: Safety & security: 80%; Research & evaluation: 20%
- AI Evaluation: Research & evaluation: 100%
- AI Governance: Society & policy: 100%
- AI Models: Models & learning: 100%
- AI Policy: Society & policy: 100%
- AI Safety: Safety & security: 100%
- Model Evaluation: Research & evaluation: 100%
- Recursive Self Improvement: Models & learning: 60%; Research & evaluation: 40%
- Superintelligence: Models & learning: 50%; Safety & security: 30%; Society & policy: 20%
Videos with Jacob Coxon
Share this person
Jacob Coxon is an AI researcher who worked on pretraining at OpenAI and Anthropic. He left Anthropic in 2026 while warning publicly about risks from the frontier AI race.