Jailbreaking Lead, Red Team

• Lead the jailbreaking practice as senior technical owner, developing universal jailbreaks and setting standards for the team.
• Develop jailbreaks against frontier models in CBRNE, cyber, and emerging domains by dismantling defense-in-depth stacks.
• Invent novel attack classes and incorporate state-of-the-art methods as frontier defences evolve.
• Mentor team members on attack craft through technical collaboration and knowledge sharing.
• Translate jailbreak findings into real-world impact by working with frontier labs and governments.

This role directly contributes to AI safety by identifying vulnerabilities in frontier models, helping to prevent catastrophic misuse in areas like biosecurity and cybersecurity.

• Senior-level expertise in AI safety and red teaming, with a focus on jailbreaking frontier models.
• Deep technical knowledge of machine learning, adversarial attacks, and defense mechanisms.
• Experience in CBRNE, cyber, or emerging domains is highly valued.
• Strong mentorship and collaboration skills.

Compensation and benefits are competitive and commensurate with experience. Specific details not provided in the posting.

FAR AI is an organization focused on AI safety research, working to ensure that advanced AI systems are developed and deployed responsibly.