OpenAI
Red Team - Cyber
$198K–$320K
San Francisco +2 · Hybrid · Full-time · Posted Aug 18, 2026
- Location
- San Francisco
Seattle
Washington, DC
About the job
Evaluates AI models’ cyber capabilities and safeguards through expert testing and automated evaluations, assesses real-world risk, and recommends mitigations.
What you'll do
- Design and run evaluations of model cyber capabilities, safeguards, and resistance to adversarial techniques.
- Build automated testing infrastructure and test cyber and agentic-system abuse risks.
- Assess and communicate findings and actionable recommendations to cross-functional partners.
What they ask for
- Substantial depth in cybersecurity or AI model evaluation.
- Working literacy across cybersecurity and model evaluation.
- Ability to write code and build practical testing tools.
- Clear communication and experience working across technical and non-technical teams.
Pay
$198K – $320K USD base salary
AI focus
Evaluate AI models’ cyber capabilities and safeguards, including policy adherence, refusal behavior, and resilience to jailbreaking; test agentic systems for abuse risks such as indirect prompt injection and agent hijacking.
Model-written summary · the original posting is authoritative