Anthropic is looking for an Anthropic AI Safety Fellow to conduct hands-on, empirical research aimed at reducing catastrophic risks from advanced AI. This four-month fellowship provides direct mentorship from Anthropic researchers, the opportunity to work on a project aligned with our safety priorities, and the goal of producing a public output like a paper submission.

What You'll Do

Work on an empirical project aligned with Anthropic's AI safety research priorities.
Use external infrastructure, such as open-source models and public APIs, to conduct research.
Produce a public output, such as a paper submission.

What We're Looking For

Motivation to reduce catastrophic risks from advanced AI systems.
Excitement to transition into full-time empirical AI safety research and interest in a full-time role at Anthropic.
A strong technical background in computer science, mathematics, physics, cybersecurity, or related fields.
Ability to thrive in fast-paced, collaborative environments.
Capacity to implement ideas quickly and communicate clearly.
Fluency in Python programming.
Availability to work full-time on the Fellows program for 4 months.
Work authorization in the US, UK, or Canada and location in that country during the program.

Nice to Have

Experience with empirical ML research projects.
Experience working with Large Language Models.
Experience in one of our core research areas, such as Scalable Oversight or Adversarial Robustness.
Experience with deep learning frameworks and experiment management.
A track record of open-source contributions.

Technical Stack

Python
Open-source models
Public APIs
Deep learning frameworks

Team & Environment

You will receive direct mentorship from Anthropic researchers. Our culture is extremely collaborative, with frequent research discussions, and we work as a single cohesive team on large-scale research efforts.