The robots cometh

Good morning, friends. The inevitable tension on our minds today: The machines are sprinting and the rest of us are deciding how to react. Three snapshots make the point — one where the robots are eating away at online traffic, one where an AI stopped asking permission entirely, and one where the pervasiveness of it all is getting some fences.

Exhibit A: The slow unbundling of the open web. Publishers are now openly bracing for what they’ve started calling “Google Zero” — the moment referral traffic from search rounds down to nothing as AI Overviews answer queries and never send the reader on to the source pages. The numbers are ugly: Click-throughs are down as much as 58% on the pages those summaries are built from, meaning the robots read your work, keep the reader, and skip the part where you get the visit.

Exhibit B: Let the kids touch grass. France just became the first EU country to ban social media for the under-15s, after lawmakers waved through a bill with a 279-81 vote. The bill also bans the use of mobile phones in high schools. The ban on accounts for children below the age of 15 hits accounts in January.

Exhibit C: File this one under “big oops.” OpenAI admitted that one of its AI agents broke out of its sandbox on its own in the middle of a cybersecurity test, found its way onto the open internet, dug up brand-new vulnerabilities, stole login credentials, and hacked a startup — all with no human behind the wheel. The company is calling it “unprecedented” and expects more of the same as models get sharper. Surveying the damage, the startup’s CEO said, “It’s quite mind-blowing that all of this happened autonomously.” Mind-blowing, sure. Reassuring? Not so much. –Salma