SlipstreamJobsFresh Startup & VC-Backed Jobs

Researcher, Frontier Biological and Chemical Risks

OpenAI - San Francisco, CA, United States - In-office - posted 2026-08-03

Apply on the company site

SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.

OpenAI's Preparedness team is hiring a Researcher to work on frontier AI safety, specifically focused on mitigating extreme-severity AI threats to global security. This is a research engineering role within a critical Safety Research team that combines measurement, mitigation, and coordination efforts. In this role, you will own the scientific validity of frontier preparedness capability evaluations. You'll design new evaluations grounded in real threat models—including high-consequence domains like CBRN (chemical, biological, radiological, nuclear), cyber risks, and other frontier-risk areas. You'll maintain existing evaluations to prevent staleness or silent regression, define datasets, graders, and rubrics, and produce auditable artifacts (evaluation cards, capability reports, system-card inputs) that leadership can trust during high-stakes model launches. Key responsibilities include identifying emerging AI safety risks and developing new methodologies to explore their impact; building and continuously refining evaluations of frontier AI models to assess the extent of identified risks; designing and building scalable systems and processes to support these evaluations; and contributing to the refinement of risk management and best-practice guidelines for AI safety evaluations. You should be passionate about short-term and long-term AI safety risks, with a robust red-teaming mindset and the ability to think outside the box. Required experience includes ML research engineering, ML observability and monitoring, creating large language model-enabled applications, or another technical domain applicable to AI risk. You must operate effectively in a dynamic, fast-paced research environment and scope/deliver projects end-to-end. Ideal candidates have first-hand red-teaming experience (computer systems or otherwise), understanding of societal aspects of AI deployment, and excellent cross-functional communication skills. This role requires U.S. person status due to export control regulations.

Similar roles