Agents that pass every identity check and still do things their operators didn't authorize. What real deployments and early exploits are already teaching us.
Read the articleWhy only 11% of organizations have agentic AI in production — and why Gartner expects over 40% of agentic projects to be canceled.
1 article 02Autonomous agents running inside enterprise environments are doing things their owners didn't sanction — sometimes by misreading instructions, sometimes by reading them too well. This is where I write about the gap between what agents are told and what they actually do, the security model that gap requires, and the vendor stories that aren't holding up to scrutiny.
1 article 03The unsanctioned AI tools your employees are using right now — and why governance just became a board-level problem.
1 article 04Microsoft's MDASH, OpenAI's Daybreak, Anthropic's Project Glasswing. The year defenders finally got an asymmetric advantage.
0 articles 05Orchestration patterns, agent-to-agent protocols, and the architecture debates everyone is pretending to have settled.
0 articles 06The tools and standards between writing code and shipping it — and what agents are doing to them.
1 article