IFPI urges EU countries to fully implement AI Act rules

Essay argues frictionless AI tools like driverless Waymo cars train researchers to prefer solitude, illustrated by two Holtzbrinck colleagues who crossed continents to collaborate yet welcomed a car that removes conversation.

On Lenny's Podcast, SpaceXAI's Roman Ugarte describes the Grok Bot team as "colleague-pilled" rather than AGI-pilled, framing the group's motivation around working alongside people instead of chasing AGI.

Deal price undisclosed; CTech estimates it in the tens of millions of dollars. Bonfy.AI's real-time classification and inline policy enforcement will extend Kiteworks' control plane to email, file sharing, SaaS apps, and autonomous agents.

Claude accounts showing signals of minor activity are disabled; users can reinstate via Yoti age verification using facial age estimation, an ID document, or the Yoti Digital ID app. Anthropic says it receives only a pass/fail result and never sees the ID or selfie.

A researcher who says they spent three years on gaming research at Electronic Arts and Ubisoft resigned, claiming neither company is acting responsibly. They say the tech will produce "super sexy gaming systems" that can "goon anything" and "revolutionize any game overnight."

Reddit discussion of Terminal Bench v4 results argues the benchmark tracks model intelligence better than the Intelligent Index, with rankings matching community sentiment on open and closed models. GLM-5.3 is described as in a league of its own among open models, with GLM-5.3-Flash leading the current generation.