How AI Agents Are Driving New Risks in Identity Security

BeyondTrust CEO Janine Seebeck on Managing Privilege in AI-Powered Environments Michael Novinson (MichaelNovinson) • December 30, 2025

Rare is the large organization that can say with certainty it knows all the artificial intelligence agents its branches have deployed - meaning organizations lack visibility into what those agents do, what systems they touch and what privileges they're inherited.

See Also: AI Arms Cybercriminals, and Defenders Must Match Pace

Granting agents access that was previously only scoped for humans leads to privilege escalation risks, since these agents interact across systems in ways that weren't foreseen, said BeyondTrust CEO Janine Seebeck. To mitigate this, organizations need insight into the roles these agents are playing, the actions they are performing and whether any of those actions violate least privilege principles, Seebeck said (see: AI as the New Middleware in Cybersecurity).

"A lot of what we're doing is try to connect those workflows, connect ...