22 Nations Demand Global AI Oversight Rules

Twenty-two countries, including France, Germany, and Japan, signed a joint call this week for coordinated global rules on artificial intelligence, arguing that no single government can manage a technology spreading this fast on its own. The timing isn't incidental. Tim Berners-Lee first sketched out a web run by autonomous software agents back in 2001, a decade after he built the thing in the first place.

For years that vision sat on the shelf as a thought experiment. Now it's showing up in products that read your email, order your groceries, and move money out of your bank account without you clicking a single button.

That shift, the one tech reporters are calling the agentic web, is forcing the question these 22 nations are trying to answer: who's accountable when software makes decisions on a person's behalf at scale. MIT Technology Review's Grace Huckins has been tracking how AI agents are rewiring the basic mechanics of the internet, and what might get lost once humans stop being the ones doing the clicking. The regulatory push and the technology are arriving on roughly the same timeline, which is either good planning or a coincidence nobody should count on.