External Researcher Access Program

External Researcher Access Program

Provides external researcher access to Anthropic systems for scientific collaboration.

The External Researcher Access Program is a rolling, in-kind grant offered by Anthropic to lower the barrier to entry for researchers working on AI safety and alignment topics. Approved applicants receive $1,000 in Claude API credits allocated to a single Claude Console organization ID, with rare exceptions for higher amounts. The award covers access to standard Claude models through the API only; it does not unlock the Claude web application, non-public or experimental models, or fine-tuning capabilities. Credits must be used in compliance with Anthropic's Usage Policy, and researchers conducting red-teaming work are not exempt from standard enforcement procedures.

Eligibility is broad: individuals, universities, research organizations, nonprofits, and for-profit entities may all apply, provided Claude API access is available in their country. For research programs involving multiple collaborators, Anthropic recommends establishing a single shared account that can distribute API access internally. Geographic restrictions apply in jurisdictions where Claude is not officially supported. Applications are submitted via a Google Form, and Anthropic evaluates all submissions on the first Monday of each month; silence after evaluation indicates non-selection.

Because the award is in-kind API credits rather than cash, it is best suited to individual researchers or small teams who need computational access to Claude for safety-focused empirical work — interpretability studies, alignment evaluations, red-teaming, or behavioral analysis — and who can operate within the standard API feature set. Groups needing larger compute budgets, fine-tuning access, or pre-release model access should consider the Anthropic AI for Science Program or the Fellows Program instead.