Researcher, Automated Red Teaming

Responsibilities

• Own research and technical direction for automated red teaming across catastrophic risk areas, initially emphasizing:
  • Automated classifier jailbreak discovery (cyber and bio)
  • Automated bio threat-development elicitation (worst-feasible planning uplift)
  • CoT monitoring evasion probing (and adjacent loss-of-control evaluations)
• Partner with vertical risk teams (Cyber, Bio, Loss of Control) to define threat models, prioritize targets, and land mitigations
• Collaborate with Classifiers team to turn attacks into training data, evals, and robustness gains
• Work with product/eng/safety stakeholders to ensure ART outputs are operationally useful

Requirements

• Strong motivation toward AI safety and reducing catastrophic risk
• Experience breaking systems and finding high-severity failure modes
• Strong applied research instincts for reproducible, interpretable evaluations
• Hands-on experience with LLMs and agents (multi-turn behaviors, tool use, adaptation to constraints)
• Ability to build scalable automation and production pipelines
• Solid software engineering fundamentals (data structures, algorithms, testing)
• Threat modeling and incentive thinking
• Clear communication to drive alignment and prioritization

Nice to have: Experience in adversarial ML, security research/red teaming, abuse prevention, or large-scale eval infrastructure.