Anthropic's mission is to create reliable, interpretable and steerable AI systems. The Anthropic Fellows Program fosters AI research and engineering talent by providing funding and mentorship to promising technical talent, regardless of previous experience. Fellows primarily use external infrastructure (open-source models, public APIs) to work on an empirical project aligned with Anthropic's research priorities, aiming to produce a public output such as a paper. The program runs 4 months full-time, with shared workspaces in London or Berkeley and remote fellows accepted in the UK, US or Canada. This workstream is AI Safety, with research areas such as scalable oversight, adversarial robustness and AI control, model organisms of misalignment, mechanistic interpretability and AI welfare.
Stack
Responsibilities
- ▹Work on an empirical AI safety research project with the goal of producing a public output (e.g. a paper)
- ▹Pursue research areas such as scalable oversight - keeping highly capable models helpful and honest
- ▹Adversarial robustness and AI control - keeping advanced AI systems safe in unfamiliar or adversarial scenarios
- ▹Model organisms - creating model organisms of misalignment to understand how alignment failures arise
- ▹Model internals / mechanistic interpretability - understanding the internal workings of large language models
- ▹AI welfare - improving understanding of potential AI welfare and developing related evaluations and mitigations
Requirements
- ▹Fluent in Python programming
- ▹Available to work full-time on the Fellows program
- ▹Strong technical background in computer science, mathematics or physics
- ▹Able to implement ideas quickly and communicate clearly
- ▹Motivated by making sure AI is safe and beneficial for society as a whole
- ▹Excited to transition into empirical AI research and interested in a full-time role at Anthropic
- ▹Work authorization in the US, UK or Canada and located in that country during the program (no visa sponsorship for fellows)
Nice to have
- ▹Motivated by reducing catastrophic risks from advanced AI systems
- ▹Experience with empirical ML research projects
- ▹Experience working with large language models
- ▹Experience in one of the listed research areas
- ▹Track record of open-source contributions
Soft skills
What we offer
- ▹Weekly stipend of 3,850 USD / 2,310 GBP / 4,300 CAD + benefits (vary by country), 40 hours/week for 4 months
- ▹Direct mentorship from Anthropic researchers
- ▹Access to a shared workspace in Berkeley, California or London, UK
- ▹Funding for compute (~$15k/month) and other research expenses
- ▹Connection to the broader AI safety and security research community