Cisco is sounding the alarm on a new class of security threat: AI agents that operate without formal oversight. In a recent interview, the company's vice president of product for identity claimed that 'shadow agents'—autonomous AI tools running outside approved governance—are present in almost every environment Cisco examines. The statement underscores how quickly agentic AI has outpaced traditional identity and access controls.
To address this, Cisco is pushing its Duo Agentic Identity product, which aims to give security teams visibility into which agents are running, what they can access, and how they authenticate. The product is positioned as a response to a market where existing identity tools were built for human users, not for software that acts on its own.
The source does not provide independent data or customer testimonials, so the claim about 'almost every environment' is Cisco's own assessment. Still, the warning aligns with broader industry concerns about AI agents making unauthorized decisions or accessing sensitive data. For businesses, the takeaway is clear: if you are deploying agentic AI, you may already have shadow agents—and you likely cannot see them.