Anthropic fellowships open now
Anthropic is recruiting for a paid Fellows Program: a four‑month research fellowship that includes a $3,850-per-week stipend plus access to roughly $15,000/month in compute credit, and it’s open to early‑career applicants with relevant work authorization. (x.com). The application window closes on April 26, making this a near-term paid route into industry alignment work. (x.com)
Anthropic is taking applications for its next Fellows cohort, a four-month paid research program that starts July 20 and has an April 26 deadline for that intake. (anthropic.com) The company says fellows will work full time for four months, get direct mentorship from Anthropic researchers, and receive a weekly stipend of 3,850 United States dollars, 2,310 British pounds, or 4,300 Canadian dollars, depending on location. (anthropic.com) Anthropic says fellows also get about 15,000 United States dollars a month for compute and other research expenses, plus access to shared workspaces in Berkeley, California, or London, with remote participation available in the United States, the United Kingdom, and Canada. (anthropic.com) The program is aimed at people trying to move into AI safety and security research, the part of the field focused on testing whether advanced systems can be controlled, interpreted, and kept from causing harm. Anthropic says fellows use open-source models and public application programming interfaces to produce a public output such as a paper submission. (anthropic.com) Anthropic is expanding the program after an earlier AI safety research cohort. The company says more than 80 percent of fellows in its first cohort produced papers, and more than 40 percent later joined Anthropic full time. (alignment.anthropic.com) This year’s workstreams are broader than the pilot. Anthropic says fellows may work on scalable oversight, adversarial robustness and AI control, model organisms, mechanistic interpretability, AI security, and model welfare. (alignment.anthropic.com) That marks a change from the 2024 pilot, which Anthropic introduced as a six-month program for 10 to 15 fellows focused on AI safety research. In that first version, the stipend was 2,100 United States dollars a week and the compute budget was about 10,000 United States dollars per fellow per month. (alignment.anthropic.com) The current posting says applicants need work authorization in the United States, the United Kingdom, or Canada and must be located in that country during the program. Anthropic says visa sponsorship is not available through this fellowship route. (job-boards.greenhouse.io) Anthropic says it will keep accepting applications for later cohorts on a rolling basis, but April 26 is the cutoff for the group that starts on July 20. For early-career researchers trying to get into industry alignment work this spring, that makes the near-term decision date unusually concrete. (anthropic.com)