Program Officer, AI Model Safety

OpenAI Foundation•San Francisco, CA

4d•$220,000 - $300,000

About The Position

OpenAI Foundation is seeking a founding grantmaker for its AI Model Safety program. This role involves turning technical expertise in frontier model evaluation, auditing, and safety into a grant portfolio and strengthening the AI safety field. The Program Officer will manage grantmaking end-to-end, from sourcing to disbursing funds, and proactively identify and help establish new organizations and initiatives. They will also develop program strategy, collaborate with the AI Resources team, and build relationships across the AI safety ecosystem. This is an early-stage, independent role reporting to the Director of AI Resources, open to mid-career and senior candidates.

Requirements

At least 3 years of experience in AI safety, ideally including hands-on research in model safety or evaluations, whether in a research role, on a safety or evals team, in grantmaking, or helping start an organization.
Deep, current fluency in AI model safety: evaluations, red-teaming, auditing, interpretability, control, and alignment research.
Ability to follow frontier AI capabilities and risks closely enough to have independent judgment about them, with views on the risks that matter most and on what each part of the ecosystem should do.
A view on where AI is going, which risks it brings, and what to do about them: your own takes on how capabilities will develop, and what that means for the safety work that will be needed.
Excel at owning ambitious, loosely scoped projects: you find the right people, convince them to start new work, and support their success.
A relentless prioritization mindset: you evaluate opportunities on impact, tractability, and neglectedness, and act on the ranking.
Ability to write and communicate compelling, transparent documents that explain your reasoning and appropriately highlight risks, uncertainties, and downsides internally.
Deeply mission-aligned and excited to help ensure AGI benefits all of humanity.

Nice To Haves

Grantmaking, VC, or philanthropic experience, especially seeding organizations or sub-fields.
A research track record in evaluations, interpretability, control, or adjacent technical safety areas.
Existing relationships across safety organizations, AI safety institutes, and AI company safety teams.

Responsibilities

Run the program's grantmaking across its domains (independent testing and evaluations, industry standards, foundational safety research, AI control, and other new ideas) from sourcing through investigation to getting resources into grantees' hands.
Run grant investigations and make grant recommendations to the board.
Proactively identify the most important projects and organizations that need to exist, and make them happen: scope priority projects, find or develop the right founders, and actively build new initiatives, seeding them if required.
Work with the AI Resources team to route AI resources to safety grantees and help them absorb it productively.
Develop the program's strategy with the team lead: what we should fund, what others in the ecosystem are doing, and where the gaps are.
Build and maintain relationships across the safety ecosystem: independent safety organizations, AI safety institutes, academic groups, other funders (philanthropic and VC alike), and AI companies' safety teams.