Mercor AI Safety Fund Grants
Mercor is putting $5 million into AI safety research grants for independent researchers and small teams. Apply with a short expression of interest.
Mercor AI Safety Fund Grants put $5 million toward safety research on frontier AI. The posting went up on 16 September 2026. The fund is separate from the Mercor Research Fellowship, which pays people to build benchmarks and economic studies.
The grant can cover researcher hours, API credits, stipends for events and conferences, and time from experts on Mercor's platform. Mercor also offers access to its expert network for grading, red teaming and annotation, and, after review, to its internal evaluation tools.
Mercor is most interested in misalignment, sandbox escape, evaluation awareness, interpretability, oversight and control, and red teaming methods. It says strong proposals outside that list are still welcome. Grantees are expected to publish a paper, an open dataset, a public method or a tool other people can use.
Independent researchers, academics, PhD students and small teams can apply. Applicants should have a specific, well scoped question and a background in machine learning, computer science, statistics or a nearby field. Apply with a one or two page expression of interest that covers the team, the project, the outputs and a rough timeline and budget.