Proteus demonstrates that adaptive red-teaming achieves 40-90% attack success after five rounds and bypasses even strong auditors at up to 41% joint success, revealing that static skill vetting underestimates residual risk.
AutoDAN: Generating stealthy jailbreak prompts on aligned large language models
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
years
2026 2verdicts
UNVERDICTED 2roles
background 2polarities
background 2representative citing papers
TurnGate identifies the critical turn in multi-turn dialogues where a response would complete hidden malicious intent, outperforming baselines on the new MTID dataset while keeping over-refusal low.
citing papers explorer
-
Proteus: A Self-Evolving Red Team for Agent Skill Ecosystems
Proteus demonstrates that adaptive red-teaming achieves 40-90% attack success after five rounds and bypasses even strong auditors at up to 41% joint success, revealing that static skill vetting underestimates residual risk.
-
One Turn Too Late: Response-Aware Defense Against Hidden Malicious Intent in Multi-Turn Dialogue
TurnGate identifies the critical turn in multi-turn dialogues where a response would complete hidden malicious intent, outperforming baselines on the new MTID dataset while keeping over-refusal low.