4 min read
Three things change when your AI is rated ‘critical’ — and ‘AGI’ isn’t one of them
A model can be genuinely more capable, genuinely worth gating, and still not be a general intelligence. All three are true of…
Artificial Intelligence
4 min read
Prompt guardrails are dead. The data layer won.
System prompts are suggestions, and app-layer approval breaks the moment an agent chains tool calls without returning to a human. The only…
7 min read
The Cyber Red Line Was Never a Wall. It Was a Turnstile.
OpenAI's Preparedness Framework was supposed to describe the line the company wouldn't cross. This week we learned it was never a wall—just…
8 min read
Container Escapes Are Dead. Authorization Escapes Won.
When agents compromise external platforms, everyone assumes the sandbox broke. The more common — and fixable — problem is that you gave…
6 min read
The Dangerous Part of Agent Fleets Isn’t the Agents — It’s the Space Between
One LLM with tool access has one door you can guard. A fleet of agents has a corridor of connecting doors, and…
5 min read
OpenAI Just Cut Off Cursor — and It Barely Matters
OpenAI is winding down the contract that supplies GPT models to Cursor after Elon Musk's SpaceX acquired the coding startup. The retaliation…
8 min read
Your Copilot Has Read the Salary Sheet. Purview DSPM for AI Just Put It on a Dashboard.
Microsoft flipped on AI data discovery by default, and the first thing it does is show you exactly how much of your…
5 min read
OpenAI Wants to Write the Rules It’s About to Break
OpenAI just launched AI Futures, a blog about how transformative AI will reshape power and governance. The essays are smart — but…