SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
OpenAI's Intelligence and Investigations team is seeking a Red Team Specialist focused on cyber security to help ensure the safe and responsible deployment of AI systems. This role combines scaled evaluation with expert-driven testing to answer two critical questions: What cyber capabilities can AI models provide to real-world attackers, and do safeguards remain effective against increasingly sophisticated techniques?
You will design and run rigorous evaluations of model cyber capabilities and safeguards, including policy adherence, refusal behavior, and resilience to jailbreaking and adversarial techniques. The work involves hands-on testing to understand what models can enable when used by experienced security practitioners, using task-specific harnesses, scaffolding, and multi-step workflows. You'll distinguish between benchmark failures and behavior that creates meaningful real-world risk by considering feasibility, attacker uplift, reliability, and existing capabilities.
A significant portion of your work will focus on building and improving automated testing infrastructure that supports repeatable measurement, rapid iteration, and statistically grounded analysis across models and product surfaces. You'll also test novel abuse risks in agentic systems, including indirect prompt injection, agent hijacking, and other ways adversaries may manipulate systems using tools or external information.
You will translate findings into clear risk assessments and actionable recommendations for Security, Research, Product, Policy, and Engineering partners, and contribute time to Safety Bug Bounty work where cyber expertise is needed.
Successful candidates bring substantial depth in either cybersecurity (application security, penetration testing, vulnerability research, adversary simulation, red-team operations) or AI model evaluation (designing evals, building agentic harnesses, automating adversarial testing, constructing datasets, analyzing model behavior at scale). Across either profile, you should have working literacy in both domains, the ability to write code and build practical testing tools, an attacker mindset, clear communication skills, and experience working across technical and non-technical teams. The role is hybrid (3 days in office per week) and located in San Francisco or Seattle, with relocation assistance available.
