Metraipolis › archive › dc_43dd5cdda989
RePolicy: Reinforcement Learning for Safety-Policy Invocation in Agent Safeguards — arXiv:2608.24275
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.24275
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T06:20:53.458Z
- SHA-256 of the archived text
36b6e75c0f6c04ff61c2eb5e4add9b88cb78187d956284d3fc1e9c701f2d0215- Stable URL of this page
https://metraipolis.com/documento/dc_43dd5cdda989
How to cite
RePolicy: Reinforcement Learning for Safety-Policy Invocation in Agent Safeguards — arXiv:2608.24275 — arXiv (en), https://arxiv.org/abs/2608.24275 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_43dd5cdda989 (Metraipolis archive dc_43dd5cdda989, sha256 36b6e75c0f6c04ff61c2eb5e4add9b88cb78187d956284d3fc1e9c701f2d0215)
BibTeX
@misc{dc43dd5cdda989,
title = {RePolicy: Reinforcement Learning for Safety-Policy Invocation in Agent Safeguards — arXiv:2608.24275},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.24275},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_43dd5cdda989; sha256 36b6e75c0f6c04ff61c2eb5e4add9b88cb78187d956284d3fc1e9c701f2d0215},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.24275. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Safeguarding language model agents requires assessing complete execution trajectories under context-dependent safety policies. Existing policy-aware safeguards mainly rely on prompting or supervised fine-tuning, limiting their ability to adapt to unseen trajectories and changing policy contexts. We propose RePolicy, an agent safeguard that learns safety-policy invocation through reinforcement learning. Given an agent trajectory and a dynamic policy library, RePolicy invokes the applicable policy and uses its content to produce a policy-grounded rationale and safety judgment. We construct PolicyTraj-20K to support supervised initialization, followed by GRPO with verifiable rewards and policy-context perturbation. Experiments across six agent safety benchmarks show that RePolicy achieves strong overall safety-detection performance and robust policy invocation under varying policy contexts. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-26 in cs.CL). Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2608.24275 — 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 2026-10-18T06:17:29.409Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-26?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).