Research · curated 31 Aug 2026
A Self-Evolving Multi-Agent Framework Defense against LLM Jailbreak Attacks
First reported arxiv.org
Coverage timeline
Single-source research — first reported, latest, and curated coincide.
Why it matters
Static jailbreak defenses cannot adapt to novel attack strategies, so a memory-based mechanism that generalizes learned rules across attack families and reduces attack success rates while preserving benign utility offers defenders an adaptive hardening approach for deployed LLMs.
A research paper by Tongyan Hu and Bryan Hooi (National University of Singapore) proposes a self-evolving, test-time multi-agent defense against LLM jailbreak attacks that uses a persistent cross-interaction rule memory. When an attack succeeds, the framework abstracts the failure into a method-level rule capturing the structural attack wrapper and reuses it against future inputs, operating entirely through external memory and prompting with no parameter updates, applicable to both open-weight and black-box API models.