Anthropic Fellows Program 2026 opens
- Anthropic opened applications for its 2026 Fellows Program on May 21, offering a four-month, full-time AI safety research fellowship with funding and mentorship. (alignment.anthropic.com) - The clearest detail is the pay package: fellows receive a weekly stipend of $3,850 in the U.S. plus about $15,000 monthly for compute. (job-boards.greenhouse.io) - Applications are rolling for cohorts starting in July 2026 and later, with another cohort expected to begin in late September. (job-boards.greenhouse.io)
Anthropic has opened applications for its 2026 Fellows Program, widening a recruiting channel the company says is aimed at early-career AI safety and research talent. The program offers four months of full-time research, Anthropic mentorship and funding support for empirical projects tied to the company’s research priorities. (alignment.anthropic.com) Anthropic said fellows will work mainly with external infrastructure such as open-source models and public APIs, with the goal of producing a public output such as a paper. (job-boards.greenhouse.io) Applications for the latest cohorts were announced on May 21. ### Who is Anthropic trying to recruit through this program? (job-boards.greenhouse.io) Anthropic says the fellowship is designed for “promising technical talent” regardless of previous experience, a formulation that broadens the pool beyond candidates with conventional AI lab resumes. The company’s program materials say it is seeking engineers and researchers to work on high-priority safety questions with direct mentor support. Jan Leike, quoted in Anthropic’s program announcement, called it “an exceptional opportunity to join AI safety research” and said fellows would collaborate with leading researchers on what he described as one of the field’s most pressing problems. (alignment.anthropic.com) Anthropic’s 2026 announcement says this year’s work will span a wider range of areas including scalable oversight, adversarial robustness and AI control, model organisms, mechanistic interpretability, AI security and model welfare. ### What does the fellowship actually include? The fellowship runs for four months on a full-time schedule, according to Anthropic’s application materials. (alignment.anthropic.com) The company says fellows receive direct mentorship from Anthropic researchers, access to a shared workspace in Berkeley, California, or London, and links to the broader AI safety and security research community. The funding terms are more specific on the application page than in secondary postings. Anthropic says fellows in the United States would receive a base stipend of $3,850 per week, with equivalent rates of 2,310 pounds in Britain and 4,300 Canadian dollars in Canada, plus country-specific benefits. (alignment.anthropic.com) The company also says it provides about $15,000 a month for compute and other research expenses. ### What kinds of projects have fellows worked on before? Anthropic says more than 80% of fellows in an earlier cohort produced papers. The company cites prior work on agentic misalignment, subliminal learning, rapid response to new ASL3 jailbreaks and open-source circuits. (job-boards.greenhouse.io) In one example, Anthropic said security fellows built agents that identified $4.6 million in blockchain smart-contract vulnerabilities and found two zero-day vulnerabilities. In another, the company said a fellow developed a rapid-response method for new ASL3 jailbreaks that later became part of Anthropic’s deployment safeguards. (job-boards.greenhouse.io) ### How closely does this connect to Anthropic’s own hiring pipeline? Anthropic says over 40% of fellows from its first cohort later joined the company full-time. That figure, published in the program announcement, shows the fellowship functioning not only as a research program but also as a route into Anthropic’s broader research and engineering teams. (alignment.anthropic.com) The company has also expanded the fellowship across teams, according to its application pages, including a dedicated AI Security workstream. Anthropic says fellows will choose and shape projects in collaboration with mentors who pitch research ideas to them. (alignment.anthropic.com) ### When do the next cohorts start, and where do applicants go? Anthropic’s main application page says it is accepting applications on a rolling basis for cohorts starting in July 2026 and beyond. The same page says the next cohort is expected to start in late September, with some flexibility for candidates who cannot match that timeline. (alignment.anthropic.com) The application is hosted through Anthropic’s careers system, while the May 21 announcement also appeared in secondary postings that summarized the program for outside applicants. Anthropic says candidates should expect an initial application and reference check, followed by technical assessments, interviews and a research discussion. (alignment.anthropic.com) (job-boards.greenhouse.io)