Metraipolis › archive › dc_e1f4e3e4c784
Extracting Forgotten Prompts from Targeted Unlearned Models — arXiv:2609.03662
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.03662
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T02:08:04.296Z
- SHA-256 of the archived text
ce89f64f592213493b34dc1d85be1c42707a0b2705205d9bde24fb87e33a137f- Stable URL of this page
https://metraipolis.com/documento/dc_e1f4e3e4c784
How to cite
Extracting Forgotten Prompts from Targeted Unlearned Models — arXiv:2609.03662 — arXiv (en), https://arxiv.org/abs/2609.03662 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e1f4e3e4c784 (Metraipolis archive dc_e1f4e3e4c784, sha256 ce89f64f592213493b34dc1d85be1c42707a0b2705205d9bde24fb87e33a137f)
BibTeX
@misc{dce1f4e3e4c784,
title = {Extracting Forgotten Prompts from Targeted Unlearned Models — arXiv:2609.03662},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.03662},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e1f4e3e4c784; sha256 ce89f64f592213493b34dc1d85be1c42707a0b2705205d9bde24fb87e33a137f},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.03662. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Recent unlearning methods (e.g. NPO, DPO, LUNAR) make use of refusal alignment to suppress forgotten data. However, it has been shown that refusal responses might leave traces of unlearning, and recent attacks have been able to successfully recover some of the unlearned knowledge. In this paper, we uncover a new vulnerability. Existing attacks typically assume that the forgotten prompts are already known to the adversary and focus on recovering their answers. However, we show that the forgotten prompts themselves can be extracted by using the retained data and black-box access to the model. Our attack, Targeted Active Search (TAS), first identifies the forgotten entities by constructing canonical templates and entity pool, and selectively querying the model using the most informative template-entity pair under a limited query budget. Once the entities are identified, TAS instantiates prompt templates with those entities to probe the unlearned model and reconstruct the forgotten prompts. Experiments across three unlearning methods with three datasets and three LLMs shows that TAS recovers the forgotten entity with $100\%$ accuracy and reconstructs up to $95\%$ of forgotten prompts, all while using up to $99.7\%$ fewer queries than naive probing. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-04 in cs.LG). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.03662 — 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 2026-10-17T01:47:25.560Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-04?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).