Today in AI — 11 March 2026

Today's top AI news — curated links and commentary on the stories that matter for product builders.

·3 min read

Every major platform company landed on the same conclusion this week: AI agents need governance. Microsoft priced it at $99/month. NVIDIA open-sourced it. The Pentagon shipped an agent builder to 3 million users without waiting for anyone's permission framework.

Agent governance becomes a product category

Microsoft launched persistent agents, a centralised control plane, and a warning that ungoverned agents become "double agents." NVIDIA's counter is NemoClaw: open-source and hardware-agnostic. Two very different bets on where the value sits.

The Anthropic standoff widens

Anthropic filed two federal lawsuits challenging its "supply chain risk" designation. Over 30 employees from OpenAI and Google DeepMind, led by Jeff Dean, filed an amicus brief warning the move threatens American AI. Meanwhile, the Pentagon launched Agent Designer for 3 million personnel to build Gemini-powered agents.

Agents need their own ecosystems

Meta acquired Moltbook, where AI agents post, comment, and vote, folding the team into Superintelligence Labs. a16z's latest Top 100 confirms agents are now their own category: ChatGPT at 900M weekly users, Claude paid subs up 200%.

The foundation layer shifts

Yann LeCun left Meta to found AMI Labs, raising $1.03B for AI that understands physics. NVIDIA's GTC next week unveils Vera Rubin with 3.3x Blackwell inference. Google rolled Gemini deep into Workspace.

Choose your governance stack before it gets chosen for you.


Stay up to date

Get notified when I publish something new, and unsubscribe at any time.

More news