Lead red-teaming exercises across Generative AI models, product surfaces, and autonomous agents to uncover novel cyber misuse capabilities, jailbreaks, and emerging security risks, conducting some volume of red teaming for Frontier Safety risks or content safety as needed. Design and execute adversarial cyber testing strategies—including evaluating exploit and malware development uplift, vulnerability discovery, phishing/data exfiltration enablement, and multi-stage cyber kill chains.
Employ technical cyber grading rubrics/safety and security/safety policies to determine effectiveness of attacks and validate testing strategies. Provide technical guidance, baseline grading, and quality oversight for specialized external red-teaming vendors and extended workforce teams. Analyze, reproduce, and validate testing results, assessing real-world exploitability and capability uplift severity. Synthesize technical findings into risk insights and data-driven narratives for cross-functional stakeholders.
Job description provided by Google.
View the original posting
