Anthropic Fellows Program, AI Security

About the role#

The Anthropic Fellows Program in AI Security is a 4-month, full-time research opportunity designed to foster technical talent. Fellows work on empirical projects aligned with research priorities, with the goal of producing public outputs such as research papers. This program provides mentorship from Anthropic researchers, access to a shared workspace, and connection to the broader AI safety and security community.

What you'll do#

• Conduct full-time research for a 4-month duration.
• Collaborate with mentors and the AI safety community.
• Work on an empirical project aligned with research priorities.
• Produce public outputs, such as papers, based on your research findings.
• Utilize shared workspace facilities.

What you'll need#

• A strong technical background in Computer Science, Mathematics, Physics, Cybersecurity, or Data Science.
• Fluency in Python programming.
• Experience with empirical machine learning research projects.
• Experience in research or engineering areas related to AI security, such as vulnerability research, pentesting, or offensive security.
• Motivation to reduce catastrophic risks from advanced AI systems.

Location & details#

• Locations: This program is available remotely or on-site in California, United States, and Ontario, Canada.
• Term: Fall 2026.
• Commitment: Full-time (4 months).
• Compensation: This is a paid fellowship.
• Sponsorship: Visa sponsorship is not provided; candidates must have independent work authorization in the United States or Canada.