Jailbreaking Lead, Red Team
Far Ai · Remote (International)
About this role
**FAR.AI — Jailbreaking Lead, Red Team** **About FAR.AI** FAR.AI is a non-profit AI research institute focused on ensuring advanced AI is safe and beneficial for everyone. Since July 2022, the team has grown to 50+ staff, published 40+ academic papers, and convened leading AI safety events. FAR.AI conducts pre-deployment testing for frontier developers and supports independent evaluations for governments, including work connected to the EU AI Office. The organization also publishes the **AI Security Leaderboard** based on its red-teaming expertise. **About the Red Team** FAR.AI’s red team aims to materially raise the bar for safety and security of widely deployed, capable AI systems. The team consistently finds failures others miss—driving real mitigations—and sets standards that labs and governments can converge on. FAR.AI also advises frontier developers on mitigations, informs the public about key AI risks, and leads high-leverage government efforts. **Role: Jailbreaking Lead** You will be the senior technical owner of FAR.AI’s **jailbreaking practice**, reporting to **Kellin Pelrine** with a dotted line to **Edward Yee**. In 2026, FAR.AI is scaling impact by: - Red-teaming major frontier model releases (closed and open-weight) within days/weeks of release - Expanding government engagements and conducting pre-deployment testing with frontier labs - Deepening testing across key risk areas (e.g., CBRN, cyber, agents) and exploring new ones (e.g., AI control and alignment) - Building tools, agents, and insights that raise the global standard for red-teaming **What you’ll do** Jailbreaking is the core technical engine of the red team. As Jailbreaking Lead, you will: **1) Lead jailbreaking on highest-stakes engagements** - Personally develop **universal and near-universal jailbreaks** against frontier closed- and open-weight models across high-risk domains (CBRNE, cyber, agentic security, extreme persuasion, and emerging risks) - Systematically dismantle **defense-in-depth stacks** (e.g., input filters, refusal/safe completion, reasoning monitors, output filters, account-level moderation) - Escalate initial vulnerabilities to reveal their **most severe form**, maximizing universality, success rate, and capability of elicited outputs - Own the technical bar for **vulnerability severity and generality** on major engagements **2) Push the frontier of jailbreaking techniques** - Invent new attack classes when existing techniques fail - Monitor and rapidly incorporate state-of-the-art methods, while building a proprietary portfolio - Shape the jailbreaking research agenda with Kellin so the toolkit stays ahead of evolving defenses - Stress-test novel affordances as frontier systems evolve (agents, tool use, long context, multimodal, reasoning, etc.) **3) Raise the technical bar across the team** - Set standards for rigor, creativity, and precision - Mentor ICs via pairing sessions, retros, and internal write-ups - Review major deliverables for technical quality, severity judgment, and clarity - If on a management track: hire, manage, and grow a jailbreaking team **without sacrificing** your personal technical edge **4) Translate jailbreaks into real-world impact** - Work directly with frontier labs and government agencies so findings lead to **real mitigations** - Contribute to public reports, benchmarks, and the FAR.AI safety leaderboard - Make precise, calibrated judgments about what is universal, reliable, and realistically actiona
Listing freshness
CronJobs last confirmed this listing 1h ago. If its source stops confirming the opening for seven days, this page is removed from active inventory.