Metraipolis › archive › dc_f5c2baea9fbf

Forgotten in Weights, Recovered by Tools: Agentic Tool Unlearning for LLM Agents — arXiv:2608.21544

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.21544
License (as archived)
CC0-1.0
Archived
2026-10-18T07:42:18.687Z
SHA-256 of the archived text
afcb143708388123b551640f605e6142e9783d739405cfcddfa3061b7d257e43
Stable URL of this page
https://metraipolis.com/documento/dc_f5c2baea9fbf

How to cite

Forgotten in Weights, Recovered by Tools: Agentic Tool Unlearning for LLM Agents — arXiv:2608.21544 — arXiv (en), https://arxiv.org/abs/2608.21544 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f5c2baea9fbf (Metraipolis archive dc_f5c2baea9fbf, sha256 afcb143708388123b551640f605e6142e9783d739405cfcddfa3061b7d257e43)
BibTeX
@misc{dcf5c2baea9fbf,
  title = {Forgotten in Weights, Recovered by Tools: Agentic Tool Unlearning for LLM Agents — arXiv:2608.21544},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.21544},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f5c2baea9fbf; sha256 afcb143708388123b551640f605e6142e9783d739405cfcddfa3061b7d257e43},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.21544. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models (LLMs) are increasingly deployed as tool-augmented agents, where responses can depend on tool calls and external observations rather than model parameters alone. This creates an evaluation mismatch for LLM unlearning: previous unlearning methods may suppress direct parametric recall, but an agent can still recover the same forget target through tools such as web search, retrieval, or database lookup. We identify this failure mode as tool-mediated recovery and study agentic tool unlearning, which aims to reduce both parametric recall and tool-mediated recovery while preserving normal tool use for retained knowledge. To address this challenge, we propose Agentic Tool Unlearning (ATU), a two-stage framework. The first stage applies parametric knowledge unlearning to suppress direct recall, while the second stage performs trajectory-level reinforcement learning in simulated tool-augmented environments to penalize target-seeking tool behavior and final-answer leakage. Experiments on RWKU and MUSE across different LLM architectures show that ATU achieves a better balance between target forgetting and retained utility, making unlearning more robust under tool-augmented agent deployment.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-25 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.21544 — 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 2026-10-18T07:37:52.820Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-25?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).