Naveen Sundaresan·Oct 18Testing a Kill Switch in Practice: A Hands-On Guide for AuditorsFrameworks point to stop mechanisms in different ways; none tells an auditor how to prove one works. This guide walks through an…
Naveen Sundaresan·Oct 4AI Can Describe the World. But Is Its Knowledge Grounded in It?Understanding is not one thing. What matters is what a system’s representations are grounded in.
Naveen Sundaresan·Sep 27Intelligence Without QualifiersWhy “artificial” and “super” may be the wrong ways to talk about intelligenceA response icon2A response icon2
Naveen Sundaresan·Sep 13Testing the Kill Switch: What Happens When You Actually Try to Stop an AI Agent?Agentic AI containment needs more than architecture diagrams and policy statements. It needs tests, evidence, and independent assurance.A response icon2A response icon2
Naveen Sundaresan·Aug 30The Load-Bearing Wall: Why Agentic AI Forces Us to Rethink Identity and Access ManagementEnterprise AI is crossing an important boundary.
Naveen Sundaresan·Aug 23Your AI System Passed Every Test. Here’s Why That Wasn’t Enough.A few weeks into deployment, an agentic AI system starts closing one category of financial crime alerts as false positives much more often…
Naveen Sundaresan·Aug 8We Set AI’s Goals. Who Set Ours?AI researchers have spent the last two years taking a strange question seriously: does any of this have an inner life. Anthropic now…
Naveen Sundaresan·Jul 26Your AI Agent Has a Kill Switch. But Will It Work?Why enterprises need kill architecture, not merely kill criteria, for autonomous AI agents
Naveen Sundaresan·Jul 19Your AI Agent Doesn’t Need to Be Compromised. It Just Needs to Be Convinced.What Five Eyes, OWASP, IMDA, MAS, NIST and the EU AI Act reveal about controlling autonomous AIA response icon1A response icon1
Naveen Sundaresan·Jul 12The AI Evaluator Gap: Can Governance Keep Pace with the AI Exponential?The blocking authority gets the headlines. The audit ecosystem is the binding constraint.