Fellowship · Anthropic

Anthropic Fellows Program, AI Safety

4 months full-time empirical AI safety research with Anthropic mentorship, shared Berkeley or London workspace, stipend and compute budget

open / rolling for July 2026 and beyond hybrid remote

Program overview

Best resume match: empirical project, agent safety, scalable oversight, clear path from production AI systems into safety research

Notes

Requires separate Constellation application form in addition to Greenhouse

View official program ↗