Agentic AI is expanding attack surfaces | Ep. 247

As enterprises embrace agentic AI, the cybersecurity challenges are multiplying. In this episode of Today in Tech, host Keith Shaw sits down with Lee Klarich, Chief Product Officer at Palo Alto Networks, to explore the growing concerns around AI security — from prompt injection and rogue agents to emerging attack vectors enabled by autonomous AI systems. Lee breaks down the three core risk categories facing companies in 2025: Misuse and lack of visibility into generative AI tools New attack surfaces opened by copilots and embedded agents The looming threat of AI-powered attackers and autonomous red teaming tools They also cover: The evolution of security from generative AI to agentic AI Real-world vulnerabilities already observed in agent communication protocols (MCP) Why discovery and control are the first steps in protecting AI deployments How security teams can evolve their playbooks without slowing down innovation The optimistic case for AI as a force multiplier for defenders—not just attackers Will the defenders stay one step ahead, or are we building systems we can’t control? This episode delivers must-watch insights for CISOs, security engineers, and tech leaders navigating the next phase of AI adoption.