AI agents now operate inside enterprises with delegated authority— reading documents, querying databases, sending messages, and executing workflows across internal systems. But what happens when that delegated authority is abused?
This talk explores a new class of insider threat: autonomous software agents acting with legitimate credentials but malicious intent. We demonstrate real-world attack paths where agents exfiltrate sensitive data across tools, perform distributed brute-force attempts under a single delegation, and manipulate internal systems without triggering traditional SIEM or UEBA alerts.
The next insider threat may not be a human. It may be the agent.
