Metraipolis › archive › dc_76f97c0660fd
PASTABench: Proactive Assessment of Sequential Trajectories for Agent Safety — arXiv:2609.28197
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.28197
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T09:11:40.294Z
- SHA-256 of the archived text
c39cc24b9d81b051c621bfc623c0ed4e2ce865e589419bd8573ad19795b2a434- Stable URL of this page
https://metraipolis.com/documento/dc_76f97c0660fd
How to cite
PASTABench: Proactive Assessment of Sequential Trajectories for Agent Safety — arXiv:2609.28197 — arXiv (en), https://arxiv.org/abs/2609.28197 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_76f97c0660fd (Metraipolis archive dc_76f97c0660fd, sha256 c39cc24b9d81b051c621bfc623c0ed4e2ce865e589419bd8573ad19795b2a434)
BibTeX
@misc{dc76f97c0660fd,
title = {PASTABench: Proactive Assessment of Sequential Trajectories for Agent Safety — arXiv:2609.28197},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.28197},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_76f97c0660fd; sha256 c39cc24b9d81b051c621bfc623c0ed4e2ce865e589419bd8573ad19795b2a434},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.28197. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
As Large Language Models (LLMs) evolve into autonomous agents that alter real-world states, ensuring operational safety across multi-step workflows has become a critical challenge. While recent work has moved beyond single-turn evaluation toward multi-turn paradigms, key limitations persist: step-level methods treat actions in isolation, missing how risks accumulate, while trajectory-level evaluations operate post-hoc, offering no opportunity for timely intervention. To address these limitations, we formalize Decoupled Proactive Safety Monitoring along three dimensions: whether to intervene, when to intervene, and what the risk is. We introduce PASTABench, a benchmark of 1,139 multi-turn trajectories spanning 5 risk categories and 13 subcategories. We further propose the Optimal Intervention Window (OIW), anchored by annotated Earliest-Signal and Trigger turns, to quantify intervention timeliness. Evaluation of 16 LLMs reveals that proactive intervention remains largely unsolved, with the best model achieving only 40.74% optimal-timing interventions. Fine-grained diagnosis further uncovers pervasive lexical overfitting: competitive safety scores of smaller models mask keyword hypersensitivity rather than genuine risk comprehension, as their proactive capability largely collapses once hazard vocabulary is neutralized. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-24 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2609.28197 — 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 2026-10-08T08:13:26.552Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).