In this episode of Safe Mode, we sit down with Philip George, Executive Technical Strategist at Merlin Group to talk about the real challenges federal agencies face at the intersection of cybersecurity, AI adoption, and post-quantum cryptography. Philip breaks down the disconnect between cyber spending and mission outcomes, why rushing into AI without sound identity management and data integrity is a recipe for disaster, and what evolving federal cryptographic requirements and shortened certificate lifecycles mean for government IT. We dig into why visibility — simply knowing what's on your network — remains the most powerful defensive posture regardless of the threat, explore the tension between zero trust and agentic AI, and hear Philip's counterintuitive take that the answer to AI-driven security challenges might just be more AI, purpose-built and narrow in scope. Also, Greg sits down with Chris Townsend, Elastic’s Global VP of Public Sector, at the Elastic Public Sector Summit to unpack how agencies can operationalize data amid rising cyber threats. Townsend explains why open standards and cross-agency data sharing matter—and how agentic AI can help modernize SOC operations by prioritizing alerts and speeding response times.

In our reporter chat, Greg Otto and Derek Johnson break down the surge of AI-in-cybersecurity developments—from Anthropic’s Project Glasswing and the “too dangerous to release” Mythos model to OpenAI’s trusted-access approach—focusing on what these tools could mean for vulnerability discovery and the balance between real risk and hype.