Metraipolis › archive › dc_0a3bfbf5612e

No Concept Escapes the Audit: Auditing-Aware Unlearning for Verifiable Concept Erasure in Diffusion Models — arXiv:2610.05401

Source: arXiv
Original: https://arxiv.org/abs/2610.05401
License (as archived): CC0-1.0
Archived: 2026-10-06T05:26:54.695Z
SHA-256 of the archived text: 1b3533b051785fb2860c14a52816de8b1bb57bf5532c8be2445608e4c3d0c6ed
Stable URL of this page: https://metraipolis.com/documento/dc_0a3bfbf5612e

How to cite

No Concept Escapes the Audit: Auditing-Aware Unlearning for Verifiable Concept Erasure in Diffusion Models — arXiv:2610.05401 — arXiv, https://arxiv.org/abs/2610.05401 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0a3bfbf5612e (Metraipolis archive dc_0a3bfbf5612e, sha256 1b3533b051785fb2860c14a52816de8b1bb57bf5532c8be2445608e4c3d0c6ed)

BibTeX

@misc{dc0a3bfbf5612e,
  title = {No Concept Escapes the Audit: Auditing-Aware Unlearning for Verifiable Concept Erasure in Diffusion Models — arXiv:2610.05401},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.05401},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0a3bfbf5612e; sha256 1b3533b051785fb2860c14a52816de8b1bb57bf5532c8be2445608e4c3d0c6ed},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.05401. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Text-to-image diffusion models can generate prohibited content, which motivates concept erasure through machine unlearning. Most erasure methods intervene at the text interface, through prompt modification or localized updates to text-conditioning weights, and they are evaluated by what the model outputs for given prompts. Such evaluation cannot see what the network still encodes. Latent-space auditing, which bypasses text conditioning and probes the denoising network directly, shows that erased concepts remain recoverable from internal representations. We find that this also holds for methods built to be robust against adversarial prompts, and that the problem grows with the number of erased concepts. We propose Auditing-Aware Unlearning for Verifiable Concept Erasure in Diffusion Models (AVCE), a framework that grounds erasure in the model's latent representations. AVCE audits the embedding neighborhood of each concept and condenses the discovered vulnerable directions into an anchor at the weakest geometric point. It edits cross-attention and self-attention projections in closed form at this anchor, then fine-tunes the two pathways with pathway-level auditing losses, using orthogonal gradient projection to consolidate multiple concepts. Experiments on SD v1.5, SDXL, and Flux 1.0 across object, explicit-content, and artistic-style unlearning show that AVCE reduces attack success rates by 5.07x and improves auditing scores by 3.84x over the strongest baseline, while preserving competitive generation quality.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.05401 — 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 2026-10-06T05:24:06.021Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).