SpecGuard: Detecting Multi-Turn LLM Jailbreaks via Specification-Based Runtime Monitoring

Summary

SpecGuard is a training-free framework for monitoring multi-turn LLM interactions. It maps observations into evidence-backed facts, expresses forbidden interaction patterns as Linear Temporal Logic specifications, and compiles them into symbolic finite automata that check dialogue traces at runtime or in retrospective audits. On multi-turn safety benchmarks it consistently outperforms representative open-source guard models while keeping false-positive rates controlled.

Topics

• Prompt injection and jailbreaks

Related items

• MediumAttackers built a ‘Skynet’ AI agent that hijacks exposed models to run cyberattacksSimilar attack · CSO Online
• InfoI-HITL: Inalienable Human-in-the-LoopSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoSteering for Safety, Hijacking for Profit: Trojan Threats in the LLM Activation Steering Supply ChainSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoThe Target Agent Is Its Own Best Critic but a Poor Proposer: Self-Improved Red TeamingSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoPersonaBreak: inducing low-safety personas for automated LLM jailbreakingSimilar attack · OpenAlex (peer-reviewed AI security)