What CISOs Can Learn From Fraud Teams
SC Media, Monday, October 12th, 2026
Fraud teams' experience with agentic commerce shows CISOs how to judge intent and context after an AI agent authenticates.
Because AI agents act as proxies for people, valid credentials and authorized access don't prove that an action is legitimate.
Fraud teams in agentic commerce already check whether agents are genuine, permitted and doing what the customer intended, and this SC Media perspective urges CISOs to borrow that approach.
It recommends constraining delegated authority by task, resource and duration, tracking authority as agents call other agents, and using behavioral and contextual signals.
These recommendations align with NIST work on agent identity and OWASP's AI agent security guidance.