OpenAI announced the launch of the Safety Fellowship, a pilot program designed to support independent researchers, engineers, and practitioners in pursuing rigorous, high‑impact research on the safety and alignment of advanced AI systems. The fellowship will run from September 14, 2026 through February 5, 2027 and offers a monthly stipend, compute support, and ongoing mentorship. Fellows may work remotely or from a Berkeley workspace and are expected to produce a substantial research output—such as a paper, benchmark, or dataset—by the end of the program. The program prioritizes safety evaluation, ethics, robustness, scalable mitigations, privacy‑preserving safety methods, agentic oversight, and high‑severity misuse domains, and welcomes applicants from computer science, social science, cybersecurity, privacy, HCI, and related fields. Applications are open now and close on May 3, with successful applicants notified by July 25. Fellows will receive API credits and other resources as appropriate, but will not have internal system access. Contact openaifellows@constellation.org for questions about the application process.
Announcing the OpenAI Safety Fellowship
OpenAI has launched the Safety Fellowship, a pilot program for external researchers, engineers, and practitioners to conduct high‑impact safety and alignment research on advanced AI systems. The fellowship runs from September 14, 2026 to February 5, 2027 and offers a monthly stipend, compute support, and mentorship. Fellows can work remotely or from a Berkeley workspace and are expected to produce a substantial research output such as a paper, benchmark, or dataset. Applications open now and close May 3, with decisions announced July 25.
Why it matters
By funding independent safety research, OpenAI aims to broaden the expertise and perspectives addressing AI alignment challenges, potentially accelerating safer AI development across the industry.
Topic dossiers OpenAI