What happens when AI policy meets implementation — and when what companies say about safety, governance, and responsibility meets what they actually do. Analysis from an award-winning journalist and product strategist who has worked across media, government, and technology.
Anthropic’s AI Slowdown Plan Doesn’t Say Who Else Decides
Amodei’s plan to slow AI shows how the U.S., its allies, and China would coordinate. It doesn’t say what role other countries or communities would play.
Anthropic Downloaded 4 Million Books. Its Settlement Covers 500,000.
Anthropic downloaded 4 million unique books and settled over about 500,000. The writers left out are not disputing anything. They were never counted.
What Claude’s watermark can’t tell you
Anthropic announced this week that new Claude models will carry invisible text watermarks. The EU rule behind them asks publishers about human review and editorial responsibility, which a watermark cannot establish.
AI Companion Rules Depend on Age Checks. Australia Shows Where They Break.
China’s AI companion rules take effect July 15. The UK set 18+ for romantic chatbots. Australia shows the problem: 7 in 10 underage users beat the checks.
Legal AI’s Strongest Protections Go to the People Who Least Need Them
Harvey got AI data governance right: no training on customer data, retention set to zero days. People fighting cases without lawyers get none of it.
A Top Clinical AI Left Europe. The FDA Eased Its US Path
OpenEvidence left the EU and UK citing AI Act requirements. In January, the FDA expanded the carveout that may keep tools like it outside US regulation.
AI Comms Isn’t Built to Hear the Audience That’s Moving the Market
AI job descriptions across vertical startups still focus on press and executive visibility, not systematic user listening. That gap now has a visible cost.
Amazon Just Settled a Pollution Case AI Reports Can’t Measure
Amazon settled Oregon groundwater pollution claims for $20.5M. The AI governance question is what sustainability reports can’t measure.
Mistral Is Selling Control of the AI Stack
Mistral’s Forge platform, Accenture deal, and European data center buildout add up to a single bet: enterprises will pay for control over their AI stack.
Two Supreme Courts Finalized AI Rules for Judges. The US Hasn’t.
Philippines and Paraguay finalized binding AI rules for their judiciaries in early 2026. The US federal judiciary still operates on non-binding guidance.
AI Chatbots Are Becoming Mandated Reporters. Without the Manual.
Two governments want AI chatbots to report users. Garbarino and China’s new rule expose a mandated-reporting role built without accountability standards.
The AI Company That Isn’t an AI Company
CoreWeave signed new deals with Meta, Anthropic, and Jane Street in one week. AI governance rules still treat “AI company” as one actor. It is not.
Allbirds Pivot Shows What PBC Status Locks In
Anthropic, OpenAI, and xAI lean on PBC status as a trust signal. The Allbirds pivot shows how a simple majority vote can strip it in one motion.
AI Companies Claim Free Speech to Block Anti-Discrimination Laws
xAI sued Colorado over its AI anti-discrimination law, arguing the rules violate the First Amendment. OpenAI backed an Illinois bill limiting AI liability.
OpenAI’s Sora Launched With a Copyright Bet. It Backfired.
Sora's opt-out copyright policy shaped Disney's restrictive deal terms and narrowed OpenAI's position before the product's shutdown.
90 Countries Are Building AI Governance. The US Isn’t One of Them.
The CAIDP Index 2026 ranks 90 countries on AI governance. The US missed the top tier. Stanford shows Americans trust their government least on AI.
AI Product Claims Are Failing. The Law Already Covers It.
AI companies across security, hiring, and healthcare face lawsuits over product claims their AI couldn't back up. The risk starts in marketing.
Iran’s War Message Is Reaching Black Americans Through TikTok’s Algorithm
Iran is using AI-generated content and decades of historical engagement to reach Black Americans on TikTok during the 2026 war. A public diplomacy analysis.
The FTC Is Using a 1914 Law to Police AI. It’s Working.
The FTC and SEC are punishing misleading AI product claims with laws written decades before AI existed. The harder cases are still coming.
AI Safety Has a Business Model Problem
Anthropic held its safety red lines and got blacklisted. OpenAI signed a Pentagon deal hours later. The market is punishing the company that said no.