Metraipolis › archive › dc_1bd47eae260e

MemLeak: Cross-User Semantic Leakage in Multi-Tenant AI Agent Memory — arXiv:2610.04195

Source
arXiv
Original
https://arxiv.org/abs/2610.04195
License (as archived)
CC0-1.0
Archived
2026-10-06T05:24:22.515Z
SHA-256 of the archived text
08a80dfbcef4d4e232d705e96f9837d4c1a4a4b9fba1748ac1951123aad705c2
Stable URL of this page
https://metraipolis.com/documento/dc_1bd47eae260e

How to cite

MemLeak: Cross-User Semantic Leakage in Multi-Tenant AI Agent Memory — arXiv:2610.04195 — arXiv, https://arxiv.org/abs/2610.04195 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_1bd47eae260e (Metraipolis archive dc_1bd47eae260e, sha256 08a80dfbcef4d4e232d705e96f9837d4c1a4a4b9fba1748ac1951123aad705c2)
BibTeX
@misc{dc1bd47eae260e,
  title = {MemLeak: Cross-User Semantic Leakage in Multi-Tenant AI Agent Memory — arXiv:2610.04195},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.04195},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_1bd47eae260e; sha256 08a80dfbcef4d4e232d705e96f9837d4c1a4a4b9fba1748ac1951123aad705c2},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.04195. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Personal AI agents in enterprise multi-tenant deployments share a common vector store for long-term memory. Shared embedding spaces create a surface for cross-user memory leakage: a user's query can retrieve semantically adjacent memories belonging to another user through ordinary cosine-similarity retrieval, without any exploit. We formalize this as cross-user admissibility failure and evaluate it across six experiments, plus follow-up ablations, under both sparse (TF-IDF) and production-faithful (MiniLM-L6-v2) retrieval. Non-adversarial, incidental leakage reaches 70--100\% under pooled {same-team} retrieval; adversarially crafted memories achieve 90--100\% top-$k$ placement, exceeding weaker keyword-based attacker baselines, with score lifts of $+0.416$ to $+0.511$ under production-faithful dense retrieval (Config B); and end-to-end response contamination reaches 5.00/5 under a production retrieval path and 4.67/5 with Claude Sonnet~4.5, with contaminated responses often scoring as helpful or more helpful than clean ones, a gap validated against human judgment. Among three architectural mitigations, only hard post-retrieval ownership gating consistently restores the clean baseline (1.00/5) across {two generation models, at a measured latency overhead of roughly 1.4~ms per query.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.04195 — 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 2026-10-06T05:24:05.439Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).