Metraipolis › archive › dc_3bb5caf7acea

TRACE: Data-Free Text Reconstruction Attacks against Approximate Unlearning in LLMs — arXiv:2610.31995

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.31995
License (as archived)
CC0-1.0
Archived
2026-10-26T20:07:13.485Z
SHA-256 of the archived text
d72f968d5c12c2230a30910d5fa61acf4f4e2a7a69039ac97704d0fd086814b7
Stable URL of this page
https://metraipolis.com/documento/dc_3bb5caf7acea

How to cite

TRACE: Data-Free Text Reconstruction Attacks against Approximate Unlearning in LLMs — arXiv:2610.31995 — arXiv (en), https://arxiv.org/abs/2610.31995 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_3bb5caf7acea (Metraipolis archive dc_3bb5caf7acea, sha256 d72f968d5c12c2230a30910d5fa61acf4f4e2a7a69039ac97704d0fd086814b7)
BibTeX
@misc{dc3bb5caf7acea,
  title = {TRACE: Data-Free Text Reconstruction Attacks against Approximate Unlearning in LLMs — arXiv:2610.31995},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.31995},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_3bb5caf7acea; sha256 d72f968d5c12c2230a30910d5fa61acf4f4e2a7a69039ac97704d0fd086814b7},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.31995. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

As large language models (LLMs) face growing demands for data removal driven by regulatory compliance, copyright concerns, and privacy protection, approximate unlearning has emerged as a practical alternative to expensive full retraining. In practice, pre- and post-unlearning model snapshots are often maintained for auditing and rollback, introducing a previously overlooked side channel. This paper demonstrates that approximate unlearning in LLM can leave residual traces within model snapshots, which may be recoverable and thus enable the privacy breaches that approximate unlearning is intended to prevent. Here, we propose TRACE, a framework that reconstructs unlearned training data given access to pre- and post-unlearning model snapshots, without any data-specific prior knowledge. To achieve this, TRACE identifies a compact set of candidate tokens from embedding-level parameter differences to constrain the search space, and assembles sequences via a contrastive scoring mechanism based on the perplexity gap between the two models. A two-stage exploration-then-completion strategy enables the recovery of multiple distinct samples. Extensive experiments on various datasets, including TOFU, MUSE, and WMDP, across multiple LLMs and six representative unlearning methods, demonstrate that TRACE achieves high-fidelity reconstruction. This paper exposes significant privacy risks in current LLM approximate unlearning deployments and highlights the need for defenses against parameter-difference leakage and appropriate management of model snapshot access.

Tema degli esploratori: fiducia digitale e crittografia (elenco arXiv cs.CR, nuovi lavori del giorno).
Categorie arXiv: Cryptography and Security (cs.CR) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.31995 — 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 2026-10-26T19:48:33.254Z dalla pagina https://arxiv.org/list/cs.CR/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).