Metraipolis › archive › dc_20c7bcdeb10a
PROACT-Agent: Progressive Runtime Oversight and Active Circuit-breaking for Real-Time Safety — arXiv:2609.34415
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.34415
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T22:38:12.569Z
- SHA-256 of the archived text
9ddc3eb155d4df062b95bae53a1b3ffe3bf60649fa653594acf51cdb99c08251- Stable URL of this page
https://metraipolis.com/documento/dc_20c7bcdeb10a
How to cite
PROACT-Agent: Progressive Runtime Oversight and Active Circuit-breaking for Real-Time Safety — arXiv:2609.34415 — arXiv (en), https://arxiv.org/abs/2609.34415 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_20c7bcdeb10a (Metraipolis archive dc_20c7bcdeb10a, sha256 9ddc3eb155d4df062b95bae53a1b3ffe3bf60649fa653594acf51cdb99c08251)
BibTeX
@misc{dc20c7bcdeb10a,
title = {PROACT-Agent: Progressive Runtime Oversight and Active Circuit-breaking for Real-Time Safety — arXiv:2609.34415},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.34415},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_20c7bcdeb10a; sha256 9ddc3eb155d4df062b95bae53a1b3ffe3bf60649fa653594acf51cdb99c08251},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.34415. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
The transition from Large Language Models (LLMs) to agents shifts safety stakes from toxic text to irreversible environmental harm. While current defenses remain largely retrospective, proactive runtime intervention is bottlenecked by the lack of large-scale, causally-consistent data. We propose PROACT-Agent, a framework for synthesizing high-fidelity trajectories to enable real-time guardrails. We identify a critical "safety drift" in prior benchmarks, where lenient annotation paradigms fail to enforce temporal consistency. PROACT-Agent addresses this through: (1) Progressive Trajectory Unrolling to reveal risks hidden in long-context interactions; (2) Reasoning-Augmented Causal Rectification to enforce monotonic causal consistency; and (3) Culturally-Aware Data Localization for cross-border robustness. We introduce PROACT-Bench, a bilingual safety benchmark with 155,780 states labeled through multi-model adjudication. Evaluating updated context before the next LLM inference, the trained guard achieves 91.46% unsafe-class F1 and 90.63% exact-boundary detection under complete source holdout. In AgentDojo, it reduces non-DoS targeted attack success from 20.82% to 0.40%. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.LG). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.34415 — 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 2026-10-07T22:14:28.171Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).