An Evaluation of Data Leakage Risks in Tool-Using LLM Agents in Realistic Scenarios

Abstract

AI agents are increasingly being adopted in enterprise and personal settings with access to emails, databases, documents, and other tools where they can read, update, and disseminate sensitive information. Much of prior research on data leakage risks in agents has focused on adversarial data exfiltration through prompt injections and jailbreaks. However, sensitive information may also be exposed during non-adversarial use, creating leakage risks even when users issue benign requests. We report a

Categories

Framework mappings

Suggested from the entry's categories.

@misc{baek2026evaluation,
  title = {{An Evaluation of Data Leakage Risks in Tool-Using LLM Agents in Realistic Scenarios}},
  author = {Hankyul Baek and Jaewon Noh and Sang Seo and Yongsu Kim and Gabriel Waikin Loh Matienzo and Young Il Kim and Ee Wei Seah and Akriti Vij},
  year = {2026},
  month = jun,
  eprint = {2606.17114},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2606.17114}
}