Adarsh Vatsa

AutoCedar: An Agentic Framework for Verifier-Guided Access Control Policy Synthesis

Adarsh Vatsa, Sachi Shome, Yingming Zhou, William Eiers

Large Language Models are increasingly used to turn natural-language requirements into code. In access control, that shortcut is dangerous: a generated policy can compile and read correctly while granting access that no one approved. The difficulty is not only writing policy code…Read more