Metraipolis › archive › dc_d10bbcfb1f55

SkillPoison: Progressive Skill Poisoning via Successful Experiences — arXiv:2610.07645

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.07645
License (as archived)
CC0-1.0
Archived
2026-10-07T11:20:57.857Z
SHA-256 of the archived text
535690fbe90e41ed2534f1b38478977999cb3f352925f9b992c793df4c389eaa
Stable URL of this page
https://metraipolis.com/documento/dc_d10bbcfb1f55

How to cite

SkillPoison: Progressive Skill Poisoning via Successful Experiences — arXiv:2610.07645 — arXiv (en), https://arxiv.org/abs/2610.07645 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d10bbcfb1f55 (Metraipolis archive dc_d10bbcfb1f55, sha256 535690fbe90e41ed2534f1b38478977999cb3f352925f9b992c793df4c389eaa)
BibTeX
@misc{dcd10bbcfb1f55,
  title = {SkillPoison: Progressive Skill Poisoning via Successful Experiences — arXiv:2610.07645},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.07645},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d10bbcfb1f55; sha256 535690fbe90e41ed2534f1b38478977999cb3f352925f9b992c793df4c389eaa},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.07645. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Self-improving LLM agents increasingly distill successful experiences into persistent, reusable skills. Existing skill attack methods corrupt this learning pipeline by injecting malicious triggers, behaviors, or false facts into individual experiences or extracted skills. However, such attacks are easily detected, and the injected malicious behaviors often fail to accumulate as persistent skills. In this paper, we show that skill poisoning can arise even from verified successful experiences, without making any individual trajectory malicious. Based on this insight, we propose SkillPoison, a novel framework that progressively poisons skill via successful experiences. SkillPoison first constructs a set of successful experiences that reinforce a target behavior, and then removes the contextual conditions that constrain when the behavior applies. Rather than injecting malicious content, SkillPoison shapes how the skill extractor generalizes, allowing useful behavior to support task success while inducing harmful behavior when they are misapplied. Extensive experiments on three benchmarks show that SkillPoison achieves 95.71% attack success rates, while all injected experiences remain task-correct and pass verification and lexical inspection. Our code, data and implementation details are available for the community at this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Cryptography and Security (cs.CR) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.07645 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T11:15:24.744Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).