Anthropic launches Fellows Program with $3,850 weekly stipend

Anthropic has opened applications for its 2026 Fellows Program, a four-month research program that pays participants $3,850 a week and provides about $15,000 a month in compute funding, according to the company’s program page and job postings. The company said fellows will work full time on empirical projects tied to Anthropic’s research priorities, with mentorship from Anthropic researchers and access to shared workspaces in Berkeley, California, or London. The postings say the program is open to candidates in the United States, United Kingdom and Canada, with remote-friendly options in the United States. Anthropic said it is reviewing applications on a rolling basis. ### How much does Anthropic say it will pay fellows? Anthropic said the expected base stipend is $3,850 per week in the United States, 2,310 pounds per week in the United Kingdom and 4,300 Canadian dollars per week in Canada. The company said the role assumes 40 hours of work per week for four months, with possible extensions in some cases. The company’s main fellows posting says participants will also receive funding for compute of about $15,000 per month, along with other research expenses and country-specific benefits. Over a four-month term, that implies roughly $60,000 in compute support in addition to the stipend. That total is an inference from Anthropic’s stated monthly compute figure and four-month program length. (anthropic.com) ### What would fellows actually work on? Anthropic’s program page says fellows will spend four months on empirical research questions aligned with the company’s priorities and are expected to produce a public output such as a paper. The company said mentors propose project ideas and fellows then choose and shape projects with them. The 2026 program page names research areas including scalable oversight, adversarial robustness and AI control, model organisms, mechanistic interpretability, AI security and model welfare. (anthropic.com) Anthropic said earlier fellows worked on topics including agentic misalignment, jailbreak defenses and open-source circuits. ### Does the program require a Ph.D. or prior AI safety experience? (alignment.anthropic.com) Anthropic’s application says it provides funding and mentorship to “promising technical talent” regardless of previous experience. The posting also says candidates should apply even if they do not meet every listed qualification. The company does not frame the fellowship as requiring a Ph.D. in the materials reviewed. (alignment.anthropic.com) Instead, the posting emphasizes technical ability, empirical research and the production of a public output. ### What does Anthropic say happened in earlier cohorts? Anthropic said that in its first cohort, more than 80% of fellows produced papers. The company also said more than 40% of those fellows later joined Anthropic full time. (anthropic.com) One example on the program page says fellows developed agents that identified $4.6 million in blockchain smart-contract vulnerabilities and found two novel zero-day vulnerabilities. (anthropic.com) Anthropic said another fellow developed a method for rapid response to new ASL3 jailbreaks, which it described as a key component of its ASL3 deployment safeguards. ### When do the next cohorts start? (alignment.anthropic.com) Anthropic’s alignment site says it opened applications for cohorts beginning in May and July 2026. The current main application posting says it is for cohorts starting in July 2026 and beyond. The job listing also says applications are being accepted on a rolling basis for the next cohort, which is expected to start in late September, and that Anthropic can sometimes accommodate different start dates. (alignment.anthropic.com) Applicants are directed to the company’s fellows application page and related workstream postings, including AI security. (anthropic.com)