Metraipolis › archive › dc_3fb28a4241dd
Black-Box Membership Inference via Word-Level Probability Estimation — arXiv:2609.10611
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.10611
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T05:46:20.070Z
- SHA-256 of the archived text
0e4693a7161d69c5dbee594a5d7567fd6cd524cbe2fd69350bfddfd72e8e1997- Stable URL of this page
https://metraipolis.com/documento/dc_3fb28a4241dd
How to cite
Black-Box Membership Inference via Word-Level Probability Estimation — arXiv:2609.10611 — arXiv (en), https://arxiv.org/abs/2609.10611 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_3fb28a4241dd (Metraipolis archive dc_3fb28a4241dd, sha256 0e4693a7161d69c5dbee594a5d7567fd6cd524cbe2fd69350bfddfd72e8e1997)
BibTeX
@misc{dc3fb28a4241dd,
title = {Black-Box Membership Inference via Word-Level Probability Estimation — arXiv:2609.10611},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.10611},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_3fb28a4241dd; sha256 0e4693a7161d69c5dbee594a5d7567fd6cd524cbe2fd69350bfddfd72e8e1997},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.10611. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Membership inference attacks (MIAs) have emerged as critical tools for auditing privacy risks in large language models (LLMs), aiming to determine whether a given text was included in a model's training corpus. However, most existing MIAs require access to per-token logits or probabilities, making them inapplicable in practice to proprietary LLMs that expose only textual continuations. To address this underexplored setting, we propose Word-level Probability MIA (WPMIA), a statistically principled MIA for strict black-box privacy auditing. WPMIA estimates word-level generation probabilities via Monte Carlo sampling with local kernel smoothing, then aggregates these estimates into a sequence-level likelihood estimator. Furthermore, WPMIA constructs the likelihood conditioned on different prefixes, thereby amplifying the distributional differences between members and non-members. We evaluate WPMIA across various open-source LLMs and find that it consistently outperforms existing black-box baselines. Importantly, we also evaluate WPMIA on modern proprietary LLMs, including GPT-5-Chat, Gemini-2.5-Flash, and Claude-4.5-Haiku, achieving an average TPR@5\%FPR of 42.0 across these models. These results offer a sound foundation for future research on strict black-box membership inference. Code is available at \href{ this https URL }{ this https URL }.
Tema degli esploratori: fiducia digitale e crittografia (retro-catalogo arXiv: annunci del 2026-09-11 in cs.CR).
Categorie arXiv: Cryptography and Security (cs.CR) ; Machine Learning (cs.LG); Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2609.10611 — 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 2026-10-16T05:30:39.365Z dalla pagina https://arxiv.org/catchup/cs.CR/2026-09-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).