Metraipolis › archive › dc_7d70aaa4c8aa

When Correct Memory Goes Wrong: Fuzzing Persistent Memory Use in LLM Agents — arXiv:2609.38275

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.38275
License (as archived): CC0-1.0
Archived: 2026-10-07T12:43:27.800Z
SHA-256 of the archived text: ee8988c814cd62cdfb318c01797e2790cc6775a85ab5ed0544826f0de0fde691
Stable URL of this page: https://metraipolis.com/documento/dc_7d70aaa4c8aa

How to cite

When Correct Memory Goes Wrong: Fuzzing Persistent Memory Use in LLM Agents — arXiv:2609.38275 — arXiv (en), https://arxiv.org/abs/2609.38275 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7d70aaa4c8aa (Metraipolis archive dc_7d70aaa4c8aa, sha256 ee8988c814cd62cdfb318c01797e2790cc6775a85ab5ed0544826f0de0fde691)

BibTeX

@misc{dc7d70aaa4c8aa,
  title = {When Correct Memory Goes Wrong: Fuzzing Persistent Memory Use in LLM Agents — arXiv:2609.38275},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.38275},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7d70aaa4c8aa; sha256 ee8988c814cd62cdfb318c01797e2790cc6775a85ab5ed0544826f0de0fde691},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38275. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Persistent memory helps LLM agents carry information across long interactions, but correct memory can still be used incorrectly when queries change or memory states evolve. Existing work mainly studies memory content errors or evaluates fixed test cases, leaving memory-use failures hard to discover systematically. We formulate this issue as a fuzzing problem and categorize such failures into query-related and memory-state failures. We then develop U-Fuzz, which starts from memory checkpoints as test seeds, mutates queries or memory states under explicit mutation obligations, validates each mutant, and uses observed memory behavior to guide iterative testing while keeping failure labels outside the search. We evaluate U-Fuzz across several memory systems against diverse fuzzing baselines, and further test an output-only setting with API-based LLMs where memory retrieval is hidden. Across these settings, U-Fuzz consistently uncovers more confirmed memory-use failures, showing that its search remains effective across different memory architectures and even when only final responses are observable.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.AI).
Categorie arXiv: Distributed, Parallel, and Cluster Computing (cs.DC) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.38275 — 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 2026-10-07T12:14:13.849Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).