Researcher, Automated Red Teaming (AI Safety)

TL;DR

Researcher, Automated Red Teaming (AI Safety): Building scalable, research-driven systems to continuously discover failure modes in AI models and their mitigations with an accent on automated classifier jailbreak discovery and bio threat-development elicitation. Focus on designing reproducible and interpretable experiments, building scalable automation for red teaming, and translating findings into actionable improvements for AI safety.

Location: Onsite in San Francisco, US

Salary: $295,000–$445,000

Company

OpenAI is an AI research and deployment company focused on ensuring that general-purpose artificial intelligence benefits all of humanity.

What you will do

• Own the research and technical direction for automated red teaming across catastrophic risk areas, with an initial emphasis on classifier jailbreak and bio threat-development elicitation.
• Partner with vertical risk teams (Cyber, Bio, Loss of Control) to define threat models, prioritize targets, and land mitigations.
• Collaborate with the Classifiers team to convert discovered attacks into training data, evaluations, and measurable robustness gains.
• Engage with product, engineering, and safety stakeholders to ensure automated red teaming outputs are operationally useful.

Requirements

• Strong motivation for AI safety and reducing real-world catastrophic risk.
• Strong applied research instincts, especially in designing reproducible, interpretable evaluations.
• Hands-on experience with LLMs and agents, including multi-turn behaviors and tool use.
• Comfortable building scalable automation, transforming red-teaming ideas into continuous pipelines.
• Solid software engineering fundamentals (data structures, algorithms, testing discipline).
• Ability to think in threat models and incentives, anticipating attacker actions and system failure under pressure.
• Capacity to translate complex findings into actionable plans and drive alignment across research, engineering, product, and policy teams.
• Focus on efficiency and prioritization, willing to decline low-leverage work to maximize risk reduction.

Nice to have

• Experience in adversarial ML.
• Experience in security research/red teaming.
• Experience in abuse prevention systems.
• Experience in large-scale evaluation infrastructure.

Culture & Benefits

• Dedicated to ensuring that general-purpose artificial intelligence benefits all of humanity.
• Committed to developing AI with safety and human needs at its core.
• Values diverse perspectives, voices, and experiences.
• Equal opportunity employer with commitments to non-discrimination and reasonable accommodations for disabilities.