GenAI CBRNE Cyber Expert
Summary
Red-teams generative AI models to uncover security flaws, design adversarial prompts, and assess risks like sensitive data leaks or CBRNE attack scenarios.
- Assess model risk for sensitive information disclosure
- Classify model failure modes by severity
- Conduct multi turn model evaluations
- Create red team test suites and evaluation rubrics
- Design adversarial prompts and jailbreaks
- Develop cyber CBRNE attack scenarios
- Document findings and provide recommendations
- Execute adversarial red teaming of generative AI models
- Maintain knowledge of GenAI attack techniques and threat intelligence