Red Team Specialist - Cyber

Job description

About the Team The Intelligence and Investigations team is dedicated to ensuring the safe, responsible deployment of AI by rapidly detecting and mitigating abuse. Our team leverages the latest testing methodologies to uncover vulnerabilities and emerging threats, helping safeguard OpenAI’s products and users. We work closely with cross-functional partners across product, policy, and engineering to drive a comprehensive defense strategy against evolving adversarial challenges. About the Role As a Red Team Specialist focused on cyber, you will help answer two practical questions: What cyber capabilities can our models provide to real-world attackers, and do our safeguards remain effective when those attackers use increasingly sophisticated techniques? The role combines scaled evaluation with expert-driven testing. You may bring deeper experience in cybersecurity and use that expertise to judge whether a model’s behavior meaningfully changes attacker capability. Alternatively, you may bring deeper experience in model evaluations, automation, or agentic harnesses and apply those skills to building rigorous cyber testing. We do not expect every candidate to be equally deep in both areas, but successful candidates will have a strong foundation in one and enough fluency in the other to work effectively across the boundary. Most of your work will focus on model cyber capabilities and safeguards; you will also spend a portion of your time testing novel abuse risks in agentic systems. This role is located in San Francisco, CA or Seattle, WA. 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 rigorous evaluations of model cyber capabilities and safeguards, including policy adherence, correct refusal, over refusal, and resilience to jailbreaking and other adversarial techniques. Conduct hands-on testing to understand what models can enable when used by experienced security practitioners, including through task-specific harnesses, scaffolding, and multi-step workflows. Distinguish benchmark or policy failures from behavior that creates meaningful real-world risk, considering feasibility, attacker uplift, reliability, and the capabilities already available elsewhere. Build and improve automated testing infrastructure that supports repeatable measurement, rapid iteration, and…