Anthropic Fellows Program for AI safety research: applications open for May & July 2026
Anthropic's Fellows Program offers a mentored, four-month research opportunity for scientists and engineers to tackle key AI safety challenges through empirical projects.
Funder: Anthropic
Due Dates: May 2026 (Cohort start) | July 2026 (Cohort start)
Funding Amounts: Weekly stipend: $3,850 USD / £2,310 GBP / $4,300 CAD for 4 months; compute funding (~$15,000/month); mentorship included.
Summary: Four-month, mentored research fellowships for engineers and scientists to conduct empirical AI safety research in priority areas.
Key Information: Requires full-time work authorization and residency in US, UK, or Canada during the fellowship.
Description
The Anthropic Fellows Program is a four-month, full-time research fellowship designed to support engineers and researchers working on Anthropic’s most urgent AI safety challenges. Fellows receive direct mentorship from Anthropic experts and work on empirical projects aligned with core research priorities, including scalable oversight, adversarial robustness, model organisms, mechanistic interpretability, AI security, and model welfare. The program emphasizes producing public research outputs (such as papers) and fosters a collaborative environment where fellows help shape their projects with guidance from leading AI safety researchers. Past fellows have contributed to advances in rapid jailbreak response, tracing internal model reasoning, and studying agentic misalignment and subliminal learning.
