Metraipolis › archive › dc_2ed42366152f

Forgetting Without Restarting: Execution-State Unlearning for Stateful LLM Agents — arXiv:2609.04875

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.04875
License (as archived)
CC0-1.0
Archived
2026-10-16T22:26:45.318Z
SHA-256 of the archived text
6e9fda94cee18ccadc07780cda7441e1d7dbc5e392c868b27fc62667dfb76745
Stable URL of this page
https://metraipolis.com/documento/dc_2ed42366152f

How to cite

Forgetting Without Restarting: Execution-State Unlearning for Stateful LLM Agents — arXiv:2609.04875 — arXiv (en), https://arxiv.org/abs/2609.04875 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2ed42366152f (Metraipolis archive dc_2ed42366152f, sha256 6e9fda94cee18ccadc07780cda7441e1d7dbc5e392c868b27fc62667dfb76745)
BibTeX
@misc{dc2ed42366152f,
  title = {Forgetting Without Restarting: Execution-State Unlearning for Stateful LLM Agents — arXiv:2609.04875},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.04875},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2ed42366152f; sha256 6e9fda94cee18ccadc07780cda7441e1d7dbc5e392c868b27fc62667dfb76745},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.04875. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Long-running LLM agents are stateful: beyond the transcript they accrete compressed summaries, plaintext memory, pending tool plans, and, under every serving API, a KV cache. Yet today's "forget" operations delete a plaintext memory record and stop, leaving every artifact derived from the revoked information intact. We formalize execution-state unlearning: after a forget request, the agent must behave as if it had never observed the target. Modeling the runtime as a deterministic transition system, we prove that the pre-target trajectory prefix is shared with this counterfactual world for free, that the post-target suffix is irreducibly tainted without token-level attribution, and that exact unlearning requires at least $T-\tau+1$ recomputed transitions, where $\tau$ is the target's injection step. Provenance-Guided Selective Replay attains this bound as a cross-layer contract spanning prompt, compressed memory, and cache: a provenance graph locates the injection point, checkpoint restoration reduces to cropping the KV cache, and sanitized replay regenerates the counterfactual suffix. Audited with elicitation, stochastic, and string-free behavioral tests across three agent suites, nine baselines, and three model families, memory deletion leaves leakage unchanged, instruction-based forgetting collapses under elicitation (Leak@probes = 1.00), and source redaction still acts on a revoked preference in 80% of episodes, while selective replay is indistinguishable from a full reset at up to 9x fewer recomputed tokens.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-07 in cs.AI).
Categorie arXiv: Cryptography and Security (cs.CR) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.04875 — 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 2026-10-16T22:18:11.541Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).