๊ฐ€์ž… ํ›„ ์ดˆ๋Œ€ ๋งํฌ๋ฅผ ๊ณต์œ ํ•˜๋ฉด ๋™์˜์ƒ ์žฌ์ƒ ๋ฐ ์ดˆ๋Œ€ ๋ณด์ƒ์„ ๋ฐ›์„ ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.

cv usk
@cv_usk
AI / Software Research Notes AI Agent, LLMOps, MLOps, Software Architecture ๆŠ•็จฟใฏๅ€‹ไบบใฎๆ„่ฆ‹ใงใ™ใ€‚
๊ฐ€์ž… May 2026
286 ํŒ”๋กœ์ž‰ ์ค‘    471 ํŒฌ
TL;DR: When attacks on search agents evolve to adapt to the defenses, they cut accuracy far more than fixed attacks. Static benchmarks may be underestimating the risk. The Web Doesn't Sit Still: Adversarial Self-Evolving Attacks on Search Agents Key points ใƒป๐Ÿงช Evaluates RAG poisoning with evolving attacks instead of fixed templates ใƒป๐Ÿงฌ CREO: contrasts successful and failed reasoning traces, and an LLM crosses over and mutates strategies ใƒป๐Ÿ” A bi-level setup where the defender discloses last round's attack patterns in its prompt ใƒป๐Ÿ“‰ GLM-5.1 on MuSiQue: base Acc 82 drops to 41 under AuthChain and 16 under CREO ใƒปโš ๏ธ Mean TASR is 60.5โ€“74.6% for CREO versus about 20% for FactSwap ใƒป๐Ÿ˜ต A new metric, APR, also measures "cognitive paralysis", where agents refuse to answer after confusing contradictions ใƒป๐Ÿ›ก๏ธ Keyword and perplexity filters barely help; even the effective defense leaves large residual risk The evaluation covers multi-hop QA only, and the defender knows the strategies. Still, it suggests search-agent safety evaluation should assume adaptive attackers. #AISecurity# #SearchAgents#
๋” ๋ณด๊ธฐ