OpenAI Launches a Grant to Secure the Future of AI

OpenAI recently announced the launch of its Safety Fellowship, an innovative program aimed at encouraging research on the safety and alignment of advanced artificial intelligence systems. This program is targeted at external researchers, engineers, and practitioners who wish to explore essential safety questions for current and future AI systems.

The program will run from September 14, 2026 to February 5, 2027, providing participants with the opportunity to conduct rigorous research in priority areas such as safety evaluation, ethics, robustness, evolving mitigations, privacy-preserving safety methods, agentic oversight, and high-severity misuse domains.

Collaboration and Resources

Fellows will have the chance to collaborate closely with mentors from OpenAI and work alongside other researchers at Constellation, located in Berkeley. However, a remote work option is also available. Participants are encouraged to produce substantial research outcomes, such as papers, benchmarks, or datasets, by the end of the program.

In addition to a monthly stipend, fellows will receive computing support and ongoing mentorship. OpenAI will also provide API credits and other relevant resources, although access to internal systems will be restricted.

Criteria and Application Process

OpenAI invites applications from diverse academic and professional backgrounds, including computer science, social sciences, cybersecurity, privacy, and human-computer interaction. The focus is on research capability, technical judgment, and execution rather than specific qualifications. Candidates will need to provide letters of reference to support their application.

Applications are open now and will close on May 3. OpenAI is committed to reviewing all submissions and will inform selected candidates by July 25. For any questions regarding the application process, interested parties can contact OpenAI at openaifellows@constellation.org.