EUVIMEDEUVIMED entdecken
Crossref

The Defense Rests: A Courtroom-Inspired Multi-Agent LLM Framework for Automated Safety Case Generation

Gerhard Yu, Amanuel Aknaw, Amir Ali Mohaghegh, Yong Song (Alex) Cao, Alvine B. Belle

Proceedings of the AAAI Symposium Series · 06.11.2026 · 10 (1) · 462-471

DOI: 10.1609/aaaiss.v10i1.43440 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Symposium Series, online erschienen am 06.11.2026. Verfasst von Gerhard Yu und 4 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Safety cases are widely used to demonstrate that mission-critical systems, such as AI-enabled autonomous driving systems and unmanned aerial systems, can operate safely after their deployment. However, manually constructing safety cases is often error-prone, time-consuming, and labor-intensive, which calls for the proposal of (semi-)automated approaches to support their creation. Inspired by the legal nature of safety case construction, we formalize the automated safety case creation problem through a courtroom analogy, where stakeholders collaboratively propose, examine, criticize, review, improve, and validate safety claims and their supporting evidence. In this analogy, the safety analyst acts as a prosecutor by identifying hazards and safety concerns, while the safety assurance case developer serves as a defense attorney by constructing and substantiating safety arguments. An orchestrator acts as a judge by coordinating stakeholder interactions, and a review board serves as a jury by assessing argument completeness and persuasiveness. Based on this formalization, we propose eTrustWare, a Large Language Model (LLM)-based multi-agent framework for automated safety case generation and review. eTrustWare relies on specialized agents that respectively embody the various roles of the aforementioned stakeholders when creating and reviewing a safety case. Our experiments on several safety cases of ML-enabled autonomous driving systems demonstrate the feasibility of coordinated LLM-based agents for generating safety case drafts that can be refined by humans.

Zitieren

APA

Yu, G., Aknaw, A., Mohaghegh, A. A., Cao, Y. S. A., & Belle, A. B. (2026). The Defense Rests: A Courtroom-Inspired Multi-Agent LLM Framework for Automated Safety Case Generation. Proceedings of the AAAI Symposium Series, 10(1), 462-471. https://doi.org/10.1609/aaaiss.v10i1.43440

Vancouver

Yu G, Aknaw A, Mohaghegh AA, Cao YSA, Belle AB. The Defense Rests: A Courtroom-Inspired Multi-Agent LLM Framework for Automated Safety Case Generation. Proceedings of the AAAI Symposium Series. 2026;10(1):462-471. doi:10.1609/aaaiss.v10i1.43440

BibTeX
@article{Yu2026,
  author = {Yu, Gerhard and Aknaw, Amanuel and Mohaghegh, Amir Ali and Cao, Yong Song (Alex) and Belle, Alvine B.},
  title = {The Defense Rests: A Courtroom-Inspired Multi-Agent LLM Framework for Automated Safety Case Generation},
  journal = {Proceedings of the AAAI Symposium Series},
  year = {2026},
  volume = {10},
  number = {1},
  pages = {462--471},
  doi = {10.1609/aaaiss.v10i1.43440}
}
Kostenlos starten →