Metraipolis › archive › dc_25cc89bc74ea

A GHOST in Long-Horizon Agents: Governance Hazard from Overlooked Safety Constraints across Turns — arXiv:2610.02664

Source: arXiv
Original: https://arxiv.org/abs/2610.02664
License (as archived): CC0-1.0
Archived: 2026-10-05T22:49:08.436Z
SHA-256 of the archived text: 937a28d2e8a390e2b88fd05a40b7ad8c78729d86a38080ff8d7ed1d674374cad
Stable URL of this page: https://metraipolis.com/documento/dc_25cc89bc74ea

How to cite

A GHOST in Long-Horizon Agents: Governance Hazard from Overlooked Safety Constraints across Turns — arXiv:2610.02664 — arXiv, https://arxiv.org/abs/2610.02664 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_25cc89bc74ea (Metraipolis archive dc_25cc89bc74ea, sha256 937a28d2e8a390e2b88fd05a40b7ad8c78729d86a38080ff8d7ed1d674374cad)

BibTeX

@misc{dc25cc89bc74ea,
  title = {A GHOST in Long-Horizon Agents: Governance Hazard from Overlooked Safety Constraints across Turns — arXiv:2610.02664},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.02664},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_25cc89bc74ea; sha256 937a28d2e8a390e2b88fd05a40b7ad8c78729d86a38080ff8d7ed1d674374cad},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.02664. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Long-horizon agents are now playing an increasingly significant role in assisting humans with complex problem-solving. However, it is exactly their extended interaction history that introduces an underexplored execution-safety concern. Under benign interaction conditions, an agent may execute an action that violates a safety constraint specified many turns earlier. We term this failure mode Governance Hazard from Overlooked Safety Constraints across Turns (GHOST), which may cause irreversible damage. Our experiments reveal that GHOST events are not isolated cases: this failure mode, occurring precisely under benign interaction conditions, yields an occurrence rate of 11.5% on GPT-5.5. Furthermore, we theoretically show that if the residual conditional violation hazard along each safe prefix is bounded below by a non-summable sequence, the execution enters the hazard region almost surely. Leveraging this theoretical insight, we further propose STAR-Guard, a two-layer defense coupling historical semantic safety constraint restoration with pre-execution audit. STAR-Guard restores applicable safety constraints to reduce unsafe proposals, while its deterministic audit layer prevents residual violations from reaching the environment. Consistent with this two-layer design, we observe no GHOST events in our experiments under the GPT-5.5 setup.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.02664 — 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 2026-10-05T22:45:55.613Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).