A paid Anthropic residency is open: $350,000 to $500,000 a year for placements of six weeks to six months, applications due November 22.
A viral post this week dresses the program up with a model called "Claude Fable 5.5" and "unlimited Claude" access. Neither exists. We checked the program page and the US job posting on November 3, 2026.
What checks out:
- Pay: "$350,000 to $500,000/year pro-rated for the duration of the residency + benefits."
- Placements: six weeks, three months or six months.
- Eligibility: "You don't need a PhD, prior ML experience, or published papers" — it targets professors, technical founders and staff-level engineers moving into AI safety research.
- Visas: "We sponsor visas for placements of three months or longer" — the viral post got this right.
- Location: San Francisco preferred, remote possible for shorter placements.
What the post invents:
- No Anthropic model is called Claude Fable 5.5; the name appears on no Anthropic page.
- Model access is not unlimited: residents get frontier models plus "a dedicated compute budget for their projects."
Verdict: real program, real pay, real deadline — but the Fable 5.5 and unlimited-Claude hook is invented.
How to apply: 1) Pick the US or UK posting on Greenhouse → 2) Submit by November 22, 2026, 11:59 PM PT → 3) Offers land in January 2027.
Deadline: applications close November 22, 2026, 11:59 PM PT.
Link: Anthropic Residency
What changed
- Applications opened with separate US and UK postings on Greenhouse, checked November 3, 2026.
- Mentors include Alignment researchers Evan Hubinger, Ethan Perez and Nicholas Carlini.
What it means for you
- Six weeks at the $350,000 low end is about $40,000 pro-rated (6 of 52 weeks), plus benefits.
- Open to people who have led ambitious technical work and are motivated by AI safety; a bachelor's degree is the stated minimum.
- Remote applicants must live in a country where Anthropic can hire.