Towards trustworthy agentic AI: a comprehensive survey of safety, robustness, privacy, and system security
A new survey examines the trustworthiness of agentic AI systems, focusing on safety, robustness, privacy, and system security. The authors highlight the risks associated with these systems and propose strategies for mitigation. They also discuss the importance of consistent evaluation metrics for high-stakes deployments.
- ▪Agentic AI systems can execute complex tasks autonomously but introduce new failure modes that challenge trustworthiness.
- ▪The survey clarifies key concepts and identifies risks along the agent workflow, summarizing targeted mitigation strategies.
- ▪Open challenges include self-evolving agents, runtime monitoring, and the trust-utility trade-off.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.23989 (cs) [Submitted on 17 May 2026] Title:Towards trustworthy agentic AI: a comprehensive survey of safety, robustness, privacy, and system security Authors:Jinhu Qi, Muzhi Li, Jiahong Liu, Yuqin Shu, Dianzhi Yu, Shicheng Ma, Wenqian Cui, Yiyang Zhao, Yiyi Chen, Ruoxi Jiang, Irwin King, Zenglin Xu View a PDF of the paper titled Towards trustworthy agentic AI: a comprehensive survey of safety, robustness, privacy, and system security, by Jinhu Qi and 11 other authors View PDF HTML (experimental) Abstract:Agentic AI systems -- Large Language Models (LLMs) augmented with planning, tool use, memory, and long-horizon interactions -- can execute complex tasks autonomously, but their multi-step trajectories introduce new failure modes that…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
