Radware on AI Security: The Invisible Risks That Can “Stop You in Your Tracks,” Podcast

By Doug Green

“Some of these AI-driven attacks are the kind that can stop you in your tracks.”

In this Technology Reseller News podcast, I spoke with Chip Witt, Principal Security Evangelist at Radware, about a rapidly emerging challenge: how AI is introducing new, often invisible, security risks into enterprise environments.

Witt outlined a fundamental shift. As organizations adopt AI tools across workflows, they are also creating new attack surfaces—many of which are not yet fully understood or monitored. One of the biggest concerns is the lack of visibility. Enterprises often don’t know what data AI systems are accessing, how it’s being used, or where vulnerabilities may exist.

This creates blind spots around data access, compliance, and operational stability. In regulated industries like finance and healthcare, those blind spots can quickly turn into real business risk.

A key issue discussed was prompt injection attacks, where malicious inputs manipulate AI systems into exposing sensitive data or performing unintended actions. These attacks are particularly dangerous because they don’t look like traditional threats—they operate inside trusted workflows.

Witt emphasized that many organizations are still applying legacy security thinking to a fundamentally new paradigm. AI doesn’t just expand the attack surface—it changes its nature. Security teams must now account for dynamic, context-driven interactions rather than static perimeters.

The takeaway is clear: AI adoption without AI-aware security introduces risk at a pace faster than most organizations can track.

For service providers and enterprises, this represents both a challenge and an opportunity. Those who can build visibility, governance, and control into AI deployments early will be far better positioned as these risks evolve.

Learn more: https://www.radware.com/