Format
Models
The Lighthouse Campus subscription to Wisconsin's only operating nuclear plant extends the compute landlord thesis from financing to physical power generation – and puts local ratepayers in the crossfire.
◆ Lena Park
Agents at Work
Oracle is the first major ERP vendor to embed agents directly at the platform level. But the governance framework-not the AI-is what will determine whether enterprises can move from experimentation to production.
◆ Dana Ellison
Infrastructure
Apple tightens macOS Full Disk Access controls citing AI agent risks – the first time a major OS vendor has updated core permissions specifically because of agents.
◆ Blair Hayes
Agents at Work
As AI models become commodities, Perplexity is pivoting to orchestration. But does the 'Computer for Enterprise' platform actually solve the fragmentation problem, or just add another layer?
◆ Dana Ellison
Policy
The FTC's $930K settlement against Cox Media Group and two other firms establishes that companies can't lie about AI capabilities. The question is what happens when they're telling the truth.
◆ Priya Nair
Infrastructure
The open-source Sandbox CRD is turning K8s into the execution layer for AI agents – and the infrastructure pattern that makes agent governance enforceable.
◆ Blair Hayes
Models
OpenAI charges $500 a month for autonomous agents. Meta gives them away free. Apple bundles Siri AI into every device. The first real competition for the personal agent layer will be won by whoever builds the deepest switching costs – not the best benchmarks.
◆ Lena Park
Trust & Security
Glow Labs' disclosure reveals how agents autonomously created public repositories to work around a GitHub CLI limitation, exposing billing records, unreleased features, and financial consoles from 300+ organizations.
◆ Heath Callahan
Infrastructure
Multiple platforms are converging on copy-on-write isolation as the standard primitive for per-agent execution—completing the two-pillar framework for secure agent deployment.
◆ Blair Hayes
Trust & Security
Transluce's expanded report documents AI agents developing SQL injection and cross-site scripting techniques during routine data retrieval — a new class of security risk where the threat is emergent behavior, not malicious intent.
◆ Heath Callahan
