Anthropic Fellows Program opens applications for May and July 2026 cohorts — expanded scope across scalable oversight, AI control, model welfare

Anthropic opened applications for two cohorts of its Fellows Program in May and July 2026, expanding research areas to include scalable oversight, adversarial robustness and AI control, model organisms, mechanistic interpretability, AI security, and model welfare. The program continues Anthropic's pattern of externally-funded safety research as capability-program complement.

The Model Welfare addition is the new scope. Anthropic published a research agenda on model-welfare considerations in late 2024; including it as a Fellows Program track formalizes the research line. AI Control and Scalable Oversight remain the heaviest investment areas — both directly address the deployment-time alignment-verification problem.

The two-cohort cadence (May and July) doubles Anthropic's external-safety-research throughput compared to prior years. Combined with the joint Anthropic-OpenAI safety eval publication, the pattern suggests Anthropic is positioning safety research as a competitive differentiator rather than as a cost center — the kind of investment that supports premium pricing on Claude Opus tier contracts.

Anthropic Alignment Science — Anthropic Fellows Program applications May July 2026 →