The world woke up to war this morning: US-Israeli forces struck Iran over the weekend, Khamenei reportedly killed, the Strait of Hormuz effectively closed, oil surged 10%, and global credit markets are rattled. In the AI lane, the week's biggest story is a spectacular paradox: Trump banned all federal agencies from using Anthropic's Claude on March 1 — then the US military reportedly used it anyway during the Iran strikes. Claude hit #1 in the App Store as a result, proving controversy is the cheapest growth marketing. OpenClaw is having its breakout week: Lex Fridman just dropped his interview with creator Peter Steinberger (@steipete), @rileybrown's "200 hours testing OpenClaw" thread is going viral, a new Diffs plugin shipped, and AI agents were integrated with ROS robotics at ClawCon. The zero-click WebSocket vulnerability (CVE-2026-25253) letting malicious sites fully hijack developer agents is pushing the Docker and cloud-isolation conversation. Anthropic quietly released its full AI training library — including Claude Code and MCP courses — for free. BCG puts the agentic AI opportunity at $200B in new tech services value over five years. The enterprise wave has arrived.
The Register interviews the NanoClaw creator — a containerized fork of OpenClaw built specifically to sandbox the agent from the host machine. Comes directly in response to the security incidents following OpenClaw's explosive popularity surge. The security-first fork for developers who learned the hard way.
Axios: Trump called Anthropic "Radical Left AI" and pulled their Pentagon contract on Friday. By Sunday, Claude was #1 in the App Store. Downloads surging. The Streisand Effect is now a growth strategy.
The Guardian: Hours after Trump's executive order banning Claude from all federal agencies, military units reportedly continued using it during Operation Epic Fury. No comment from DoD or Anthropic. The policy-reality gap in AI governance laid bare.
Agentic AI dominates the March news cycle across finance, marketing, and professional services. Also: Anthropic blocked 24,000 fraudulent accounts allegedly created by Chinese AI labs to scrape and abuse Claude's API.
Zero-click commerce is the defining retail shift: shoppers may never click or visit a site to buy. BCG analysis: agentic AI could unlock $200B in new tech services value over five years. The competitive divide is now between firms that use AI to inform their people vs. firms that use AI to run their processes. If you no longer wish to receive this brief, please reply with the word "remove". Max · maximus.cahall@gmail.com · Mon, Mar 2, 2026
Selected posts from the edition, with links to the originals.
(retweeting @lexfridman): "Here's my conversation with Peter Steinberger (@steipete), creator of OpenClaw, an open-source AI agent that has taken the world by storm…" — 1,068 retweets
View post on X ↗"New in OpenClaw: Diffs plugin 🔧 Your agent can now render actual diffs — syntax-highlighted, unified or split view. One tool call away."
View post on X ↗(retweeting @vitl2907): "AI agents can now control robots! For ClawCon, we integrated @openclaw and @rosorg — the largest open-source robotics stack in the world. Your agent can now issue real-world robot commands." — 108 retweets
View post on X ↗"I spent 200 hours testing OpenClaw, trying to find the perfect setup… My biggest takeaway: Keep your agents focused, and…" — 123 retweets and climbing
View post on X ↗"Anthropic has made its full AI training library publicly available for free, with certificates. Course catalog: Claude Code in Action · Claude 101 · Intro to Model Context Protocol · MCP Advanced Topics · Claude Code in Action."
View post on X ↗"✅ ACQUIRED — AI girlfriend app. Listed on TrustMRR 16 days ago." — Marc Lou keeps acquiring solo AI apps and flipping them. Execution as a business model.
View post on X ↗"Researchers at Tsinghua University have developed 'Agent Hospital' — ~42 AI doctor agents across 21 medical specialties. The platform simulates patient cases so the AI can study disease patterns, practice diagnosis, and improve treatment."
View post on X ↗"Stanford HAI audited 28 privacy documents across the six biggest AI companies in the US. Amazon. Anthropic. Google. Meta. Microsoft. OpenAI. The results are… not great."
View post on X ↗"Deepseek just broke the one rule every transformer has followed for a decade 🤯 x + f(x) — the residual connection. If you don't understand why this matters for LLM architecture, read this."
View post on X ↗"🛡️ OpenClaw 0-Click Vulnerability Allows Malicious Websites to Hijack Developer AI Agents — A newly identified zero-click vulnerability in OpenClaw's local WebSocket gateway allows attackers to compromise developer systems with minimal interaction. No plugins. No extensions. No user action required."
View post on X ↗