Metraipolis › archive › dc_8b4516604df2

Accept Changes, Reject Attacks: Robust Memory Management for LLM Agents — arXiv:2610.41056

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.41056
License (as archived)
CC0-1.0
Archived
2026-11-02T07:30:49.820Z
SHA-256 of the archived text
dbceb91c6583bce41213f43a6920bf85b858e9b56acbe0663eb00ad0ff1f5418
Stable URL of this page
https://metraipolis.com/documento/dc_8b4516604df2

How to cite

Accept Changes, Reject Attacks: Robust Memory Management for LLM Agents — arXiv:2610.41056 — arXiv (en), https://arxiv.org/abs/2610.41056 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_8b4516604df2 (Metraipolis archive dc_8b4516604df2, sha256 dbceb91c6583bce41213f43a6920bf85b858e9b56acbe0663eb00ad0ff1f5418)
BibTeX
@misc{dc8b4516604df2,
  title = {Accept Changes, Reject Attacks: Robust Memory Management for LLM Agents — arXiv:2610.41056},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.41056},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_8b4516604df2; sha256 dbceb91c6583bce41213f43a6920bf85b858e9b56acbe0663eb00ad0ff1f5418},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.41056. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Since large language model (LLM) agents rely on long-term memory, what they store now shapes what they do later. A memory manager decides how each input changes this memory and which records the agent acts on. It can fail in two ways: keeping a fact that has since changed (outdated memory), or storing a planted passage or injected instruction (memory poisoning). Most prior work studies them separately, yet they call for opposite write decisions, so no fixed write rule handles both. We therefore formalize memory robustness as handling both without doing worse on other inputs. Since accuracy checks only the answer, we measure robustness with evidence-conditioned accuracy (ECA), which also checks that the records the manager selects include all the evidence the answer needs, updated and free of attacks. However, existing memory systems and defenses are not more robust than an untuned manager on our test sets. We improve robustness with a simple one-shot recipe that fine-tunes the manager on episodes synthesized around a single conversation with inserted changes and attacks. On three backbones, it improves every category score on our test sets and more than doubles the untuned manager's overall ECA, though robustness remains incomplete.

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/2610.41056 — 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 2026-11-02T07:26:57.789Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).