Anthropic Residency: applications open for established technical leaders

Open original

Anthropic is opening applications for a paid, six-week to six-month research residency that brings physicists, mathematicians, founders and senior engineers into full-time AI safety work, with no ML background required. Applications for this round close November 22, 2026.

  • Pay is $350,000 to $500,000 a year equivalent, and residents are mentored one-on-one by Anthropic researchers and given a dedicated compute budget on frontier models.
  • The program is built on the Anthropic Fellows model; past fellows produced work on subliminal learning, the Petri auditing tool, and persona vectors.
  • Research areas include scalable oversight, adversarial robustness, interpretability, model organisms of misalignment, AI welfare, and training infrastructure. Residents are preferred in San Francisco, though remote participation is possible for shorter placements.

Anthropic is opening applications for a paid, fixed-term research program aimed at people who already have serious technical careers but no background in machine learning. Placements run six weeks, three months, or six months, and residents are paid the equivalent of $350,000 to $500,000 a year. The target group is senior quantitative people: professors in fields like physics or mathematics, technical founders, and staff-level engineers. The pitch is that the skills that matter most in AI research are increasingly about framing a research agenda and judging what is worth doing, and those are things senior people in other fields already tend to be good at. Applications for this round close November 22, 2026, through separate US and UK postings. The program is modeled on the Anthropic Fellows Program, which the article credits with producing work on subliminal learning, the Petri auditing tool, and persona vectors. Residents get a mentor paired with them for regular one-on-one guidance, dedicated self-study time, talks to get them up to speed on modern ML and safety research, and a compute budget for working with frontier models. They are expected to work from the San Francisco offices, though remote participation is possible for shorter placements. Anthropic says the goal is for every resident to be able to continue doing AI research full-time afterward, either at Anthropic or somewhere else. The research areas listed are broad: scalable oversight, adversarial robustness and AI control, model organisms of misalignment, interpretability, safeguards, AI welfare, stress-testing frontier models in areas like cybersecurity and robotics, and the infrastructure behind training runs. The article names nine potential mentors, including Ethan Perez (Alignment Science), Evan Hubinger (Alignment Stress-Testing, formerly at MIRI), Samuel Marks (Scalable Oversight), Jack Lindsey (Model Psychology), and Nicholas Carlini (adversarial ML and AI security, formerly at DeepMind and Google Brain). The article also points readers to Anthropic's technical safety research directions document and lists representative Fellows papers, such as SHADE-Arena on sabotage and monitoring and AuditBench on hidden-behavior auditing, which give a concrete sense of the kind of work residents would be doing.

Ask about this article