Research Program Manager, Frontier Assurance

About the Team

The Frontier Assurance team brings independent scrutiny into OpenAI’s safety decisions and helps the public understand and assess our safety work. We lead third-party assessments and safeguard testing for OpenAI’s flagship launches, pilot new assurance mechanisms such as embedded auditing, run our misalignment disclosure process, and incorporate independent expert input as evidence for critical safety decisions.

About the Role

As a Research Program Manager on the Frontier Assurance team, you will build programs that bring independent expertise into frontier AI safety decisions and make the evidence behind those decisions understandable to the public. You will lead external research partnerships and third-party assessments, coordinate public safety documentation, and develop new approaches to independent scrutiny and transparency. Working across research, engineering, product, policy, and communications, you will help ensure external findings inform concrete decisions and that our public explanations accurately reflect the evidence, limitations, and remaining uncertainty.

We’re looking for people with deep experience in research partnerships and program management with technical and research teams. This role combines partnership management, cross-functional coordination, an understanding of AI safety research, alignment, and evaluations, and strong communication skills. You will work with researchers and engineers within OpenAI and across the external community to initiate projects, set ambitious goals and milestones, and drive execution across multiple teams.

This role is based in San Francisco, CA. We use a hybrid work model of 3 days in the office per week and offer relocation assistance to new employees.

In this role, you will:

• Design and run third-party assessment programs for frontier models and safeguards, including independent evaluations, adversarial testing, and new approaches such as embedded auditing. Work with researchers and external partners to define assessment questions, scope, access, timelines, and deliverables.

• Build and manage strategic research partnerships with third-party evaluators, academic labs, and other independent experts, prioritizing expertise, independence, and diversity of perspectives. Enable rigorous research, including work that challenges internal assumptions.

• Bring external findings to safety decision-makers and translate them into actionable recommendations for safeguards, deployment, product, and policy. Track follow-up actions and ensure partners understand how their input was considered.

• Partner with research, product, policy, and communications teams to explain the evidence behind OpenAI’s safety approach: what we tested, what we learned, how findings shaped safeguards and deployment decisions, and where limitations and uncertainty remain. Translate complex technical results into accurate, accessible communication for expert and public audiences.