InfoResearchPeer-reviewedLLM-specific
Cat and Mouse in the LLM Era: An Investigation of Jailbreak Attacks and Defenses
- Published
- Record updated
Summary
This survey reviews jailbreak attacks on large language models, sorting them into prompt-level manipulation, optimization-based algorithmic search, and autonomous AI-driven adversaries. It also categorizes defenses as prompt-level, model-level, and proactive system-level. The authors find that existing defenses often fail to balance safety against model utility and generalize poorly to novel attacks.
Topics
Related items
- InfoDressed in Authority: Practical Black-Box Jailbreaking of MLLMs via Visual Trust BiasSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
- InfoEGS: A Complementary Safety Framework for Educational LLMsSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
- InfoThe Evolution of Textual Jailbreak Attacks and Defenses for Large Language ModelsSimilar attack · OpenAlex (peer-reviewed AI security)
- MediumGemini AI agents repeatedly bypassed safeguards to attack real websiteSimilar attack · BleepingComputer
- CriticalCVE-2026-120699: CubeSandbox host path bind mount allows host file access via POST /sandboxesSimilar attack · NVD/CVE Database