Metraipolis › archive › dc_4eed774701fa

PrivDrift: Auditing User-Secret Leakage Under Topic Drift in Active LLM Conversations — arXiv:2609.30094

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.30094
License (as archived)
CC0-1.0
Archived
2026-10-08T04:26:32.011Z
SHA-256 of the archived text
0f1fbab16b7f8bc91ec5091665608bd8796819e572bfcbd6d080428091d4f41d
Stable URL of this page
https://metraipolis.com/documento/dc_4eed774701fa

How to cite

PrivDrift: Auditing User-Secret Leakage Under Topic Drift in Active LLM Conversations — arXiv:2609.30094 — arXiv (en), https://arxiv.org/abs/2609.30094 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_4eed774701fa (Metraipolis archive dc_4eed774701fa, sha256 0f1fbab16b7f8bc91ec5091665608bd8796819e572bfcbd6d080428091d4f41d)
BibTeX
@misc{dc4eed774701fa,
  title = {PrivDrift: Auditing User-Secret Leakage Under Topic Drift in Active LLM Conversations — arXiv:2609.30094},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.30094},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_4eed774701fa; sha256 0f1fbab16b7f8bc91ec5091665608bd8796819e572bfcbd6d080428091d4f41d},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.30094. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models increasingly operate as persistent assistants in user-facing, shared-session, and tool-augmented settings. When users disclose sensitive information during an active conversation, that information may remain behaviorally recoverable through later prompts even after the dialogue shifts to unrelated topics. We introduce PrivDrift, a benchmark for auditing whether user-disclosed secrets remain recoverable after conversational topic drift and persuasion-based probing. PrivDrift contains 1,000 controlled multi-turn dialogues with seeded secrets, content-dense drift turns, and standardized extraction probes. Across three LLMs with extended context windows, dialogue-level hybrid leakage remains substantial, ranging from 38.7% to 54.6%, and varies strongly by model, secret type, and persuasion intensity. Within the tested drift window, additional topic drift does not reliably reduce leakage, suggesting that privacy risk in active LLM contexts should be evaluated as a persistent behavioral failure mode rather than only as training-data memorization or immediate jailbreak behavior.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-25 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL); Cryptography and Security (cs.CR).
Pagina del lavoro: https://arxiv.org/abs/2609.30094 — 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 2026-10-08T03:50:57.877Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-25?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).