Built with and by Teycir Ben Soltane•
How to Use•FAQ•GitHub•arXiv.org•
Share:
ArXivCSExplorer
☆☆Bookmarks🏆RSSHow to UseFAQ
Back to Paper
cs.CRcs.AI

Local ID: 2603.28166v2

AI Summary: gemma4:e4b

Evaluating Privilege Usage of Agents with Real-World Tools

By Quan Zhang, Lianhang Fu, Lvsi Lian, Gwihwan Go, Yujue Wang, Chijin Zhou, Yu Jiang, Geguang Pu

Revision History Timeline

v13/30/2026
3/30/2026

“Accepted to the FSE 2026 Ideas, Visions, and Reflections track”

v24/20/2026
4/20/2026

“Accepted to the FSE 2026 Ideas, Visions, and Reflections track”

★ Version indexed in Explorer

Comparing v1 vs v2

Green = Added • Red = Removed

Title Comparison

Evaluating Privilege Usage of Agents onwith Real-World Tools

Authors Comparison

No author changes.

v1 Comment

“Accepted to the FSE 2026 Ideas, Visions, and Reflections track”

v2 Comment

“Accepted to the FSE 2026 Ideas, Visions, and Reflections track”

Abstract Word Diff

Equipping LLM agents with real-world tools can substantially improve productivity. However, granting agents autonomy over tool use also transfers the associated privileges to both the agent and the underlying LLM. Improper privilege usage may lead to serious consequences, including information leakage and infrastructure damage. While several benchmarks have been built to study agents' security, they often rely on pre-coded tools and restricted interaction patterns. Such crafted environments differ substantially from the real-world, making it hard to assess agents' security capabilities in critical privilege control and usage. Therefore, we propose GrantBox, a security evaluation sandbox for analyzing agent privilege usage. GrantBox automatically integrates real-world tools and allows LLM agents to invoke genuine privileges, enabling the evaluation of privilege usage under prompt injection attacks. Our results indicate that while LLMs exhibit basic security awareness and can block some direct attacks, they remain vulnerable to more sophisticated attacks, resulting in an average attack success rate of 84.80% in carefully crafted scenarios.
View Full Version History on arXiv