Anthropic Launches $150M Claude Corps for Nonprofit AI Fellows

Most AI labs measure their social impact through grants and safety papers. Anthropic is trying something structurally different: a $150 million fellowship that puts early-career workers directly inside nonprofits as full-time AI practitioners.

Claude Corps, announced by Anthropic, will train 1,000 fellows and place them in year-long, full-time positions at nonprofits across the United States. Fellows earn $85,000 annually plus benefits. The eligibility bar is intentionally low: applicants must be over 18 with fewer than two years of full-time work experience, no educational credential required, with authorization to work in the US and willingness to relocate if needed.

The program runs through a three-organization structure. Anthropic provides funding and Claude expertise. CodePath serves as the official employer and leads the programming, which includes weekly five-hour training sessions and technical office hours from Anthropic staff. Social Finance handles measurement, evaluation, and the harder job of building scalability infrastructure for future cohorts. At least 400 nonprofits are set to participate, spanning education, food security, veteran services, marine conservation, and youth development. The first cohort of 100 fellows has an application deadline of July 17, 2026, with a start date of October 2026; two additional cohorts are planned for January 2027 and August 2027.

The honest caveat is that the infrastructure to scale this beyond the first three cohorts is still being built. Social Finance's measurement work will be load-bearing: if early cohort outcomes are unclear or underwhelming, subsequent funding rounds face real headwinds before the program reaches its 1,000-fellow target. The announcement also does not address what happens to fellows after the placement year ends, or how host nonprofits are selected and held accountable for providing meaningful work.

For the nonprofits involved, the upside is direct: skilled AI practitioners at no direct cost, in sectors that have long been underserved by AI tooling. For Anthropic, 400-plus civic deployments of Claude generate the kind of real-world feedback and positioning that no benchmark can replicate.