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.
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.
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.