Metraipolis › archive › dc_ca6238051f0b

Sapien: A Stateful Policy Engine for Autonomous AI Agents — arXiv:2610.00797

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.00797
License (as archived)
CC0-1.0
Archived
2026-10-06T12:02:52.823Z
SHA-256 of the archived text
53ce8c8031e0ad8f897e332075e15fb1cf5c0cb9e3229ee2d931b7ff40a89e11
Stable URL of this page
https://metraipolis.com/documento/dc_ca6238051f0b

How to cite

Sapien: A Stateful Policy Engine for Autonomous AI Agents — arXiv:2610.00797 — arXiv (en), https://arxiv.org/abs/2610.00797 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ca6238051f0b (Metraipolis archive dc_ca6238051f0b, sha256 53ce8c8031e0ad8f897e332075e15fb1cf5c0cb9e3229ee2d931b7ff40a89e11)
BibTeX
@misc{dcca6238051f0b,
  title = {Sapien: A Stateful Policy Engine for Autonomous AI Agents — arXiv:2610.00797},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.00797},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ca6238051f0b; sha256 53ce8c8031e0ad8f897e332075e15fb1cf5c0cb9e3229ee2d931b7ff40a89e11},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.00797. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Contextual security defenses prevent AI agents from taking rogue actions by synthesizing a task-specific policy and enforcing it on the agent's tool calls. In multi-step tasks, however, which actions are valid often depends on what the agent has already done and learned. We present Sapien, a policy engine for enforcing stateful contextual policies. A Sapien policy specifies permitted tool-call sequences using a regular expression extended with stateful predicates, deferred policy generation, and scoped semantic checks. We show that Sapien stays within a few percent of an unconstrained agent's utility. Even if the agent is fully hijacked, Sapien's policies rule out 93-95% of attacks on AgentDojo and 62-85% on Toolathlon (twice as many as tool allowlists on long-horizon tasks).

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-02 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL); Cryptography and Security (cs.CR).
Pagina del lavoro: https://arxiv.org/abs/2610.00797 — 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 2026-10-06T10:21:27.889Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).