Anthropic AI Safety Fellow

Details

Anthropic’s Fellows Program accelerates AI safety research by funding and mentoring promising technical talent for a four-month empirical research project. Fellows primarily use external infrastructure (open-source models, public APIs) to conduct an empirical project aligned with Anthropic’s research priorities and aim to produce a public output (e.g., a paper submission). This application is for cohorts starting in May and July 2026. Please apply by January 12, 2026.

Responsibilities

• Conduct an empirical AI safety research project over four months using external infrastructure (open-source models, public APIs).
• Work toward producing a public research output (paper, blog post, dataset, or open-source artifact).
• Participate in mentor matching and project selection processes.
• Collaborate with Anthropic researchers and the broader AI safety research community.
• Attend weekly research discussions and check-ins with mentors.

Requirements

• Must be fluent in Python programming.
• Available to work full-time (40 hours per week) on the Fellows program for four months.
• Have work authorization and be located in the US, UK, or Canada for the duration of the program (Anthropic is not able to sponsor visas for fellows).
• Strong technical background in computer science, mathematics, physics, cybersecurity, or related fields (Bachelor’s degree or equivalent experience required for logistics).
• Thrive in fast-paced, collaborative environments and be able to implement ideas quickly and communicate clearly.

Strong candidates may also have:

• Experience with empirical ML research projects.
• Experience working with large language models (LLMs).
• Experience in AI safety research areas (scalable oversight, adversarial robustness, model organisms, mechanistic interpretability, AI welfare).
• Experience with deep learning frameworks and experiment management.
• Track record of open-source contributions.

Logistics & eligibility:

• Fellows must have work authorization in the US, UK, or Canada and be located in that country during the program.
• We are open to remote fellows in the UK, US, or Canada; shared workspaces are available in London, UK and Berkeley, California, and mentors will visit these spaces.
• Visa sponsorship is not available for fellows.

Interview process:

• Initial application & references check, technical assessments & interviews, and a research discussion.

Compensation

• Weekly stipend: 3,850 USD / 2,310 GBP / 4,300 CAN.
• Funding for compute (~$15k/month) and other research expenses.
• Expected commitment: 40 hours per week for 4 months (with possible extension).

Mentorship & Research Areas

• Direct mentorship from Anthropic researchers and access to a broader AI safety research community.
• Example mentors include Jan Leike, Sam Bowman, Nicholas Carlini, Jascha Sohl-Dickstein, and others.
• Research areas include scalable oversight, adversarial robustness and AI control, model organisms, mechanistic interpretability, and AI welfare.

Benefits

• Access to a shared workspace (London or Berkeley) and remote options in the UK, US, or Canada.
• Compute funding and research expense support.
• Connection to Anthropic researchers and potential pathways to full-time research roles (no guarantee of full-time offers).

How to Apply

• Apply through the Constellation application portal (link provided in the original posting).

Additional Notes

• Anthropic encourages applicants from diverse backgrounds and explicitly invites candidates to apply even if they do not meet every listed qualification.
• Applications are due January 12, 2026.