Anthropic Fellows Program 2027: AI Safety and Security Research Fellowship
Overview
The Anthropic Fellows Program is a four-month, full-time initiative designed for aspiring researchers and engineers who want to undertake intensive technical work in artificial intelligence safety and security. Participants engage in projects aligned with company priorities, with the objective of producing public outputs such as academic papers and other research contributions. The cohort connects participants with the wider research community and offers potential pathways toward future roles, though full-time employment is not guaranteed.
What you get
Selected participants receive a weekly stipend of US$3,850, £2,310, or CAD$4,300 depending on their location, alongside research computing funding of approximately US$15,000 per month and additional funding for approved research expenses. Fellows also benefit from direct mentorship by experienced Anthropic researchers, access to shared workspaces in London or Berkeley, opportunities to engage with the broader research community, and potential pathways to future positions.
Who can apply
The programme welcomes promising technical candidates, including individuals looking to transition into empirical AI research, with experience in machine learning, large language models, cybersecurity, or related technical disciplines. Applicants must demonstrate fluency in Python programming, be available for full-time participation requiring approximately 40 hours of work per week, and possess full-time work authorisation for the United States, United Kingdom, or Canada. Participants must be able to reside in their authorised country throughout the fellowship, as visa sponsorship is not provided.
Requirements
Applicants should have a strong technical background in fields such as computer science, mathematics, physics, or complementary areas like economics, social sciences, or cybersecurity. Candidates must submit an updated curriculum vitae highlighting Python proficiency, technical projects, research experience, and open-source contributions. The evaluation process includes reference checks, technical assessments, interviews, and a research discussion.
Key dates
The application deadline for the January cohort is 18 October 2026.