Quick Hits
Six things worth knowing from Tuesday's quieter corner of AI: a security vendor claiming frontier models of its own, a cheaper way to make Gemini watch video, and a token price index that keeps sliding. CrowdStrike opened a Cyber Superintelligence Lab and shipped SafeMind, its own security models
The Guardrails
Two moves this afternoon define where AI is actually being trusted: OpenAI wired ChatGPT straight into the electronic health record that most US hospitals run on, and the EU finalised a rulebook that treats a chatbot as infrastructure subject to public-safety law. Different rooms, same question — who carries the
The Stack
A token is the chunk of text an AI model actually reads and writes — not a word, not a character, but a piece of a word: "un", "believ", "able". Every language model, from the smallest phone-side model to the biggest frontier system, does
The Arena
Two launches and one filing, all inside a few hours: Google shipped a Canva-shaped design tool into Workspace, John Deere put an assistant on top of a decade of farm telemetry, and SoftBank's data center developer filed to go public while admitting it has no operating data
The Arena
Waymo lit up paid driverless service in Denver, San Diego and Tampa on Tuesday, taking the U.S. robotaxi count to 14 cities and more than 4,000 vehicles. The detail most coverage skipped: in two of those three cities, the only car in the fleet is the new Zeekr-
The Stack
Today's daily board has a clear pattern: agents are no longer just writing code in a terminal, they're producing finished artifacts — a paper, a cut video, a cloned voice, a model of your own. Five fresh projects, none of them repeats from this week's
The Everyday
Robotaxi season just went national. Waymo opened paid rides in three more U.S. cities today, Amazon's Zoox said it will start supervised testing in two more, and Tesla has a Cybercab event on Thursday — all inside 72 hours. Waymo is now live in 14 U.S. cities.
The Guardrails
Security validation has spent a decade proving you can block a single technique. Filigran's new release argues that's the wrong question — and puts a camera-equipped toothbrush and a very large open-source agent release on the same Tuesday. Filigran shipped OpenAEV v3, and the headline
The Guardrails
Sony Music Publishing and Warner Chappell Music filed a sweeping federal copyright lawsuit against Anthropic late Friday, accusing the AI lab of running a "brazen campaign of illegally torrenting, scraping, and downloading" tens of thousands of songs to train Claude — and they put CEO Dario Amodei and co-
The Arena
Google is doing something it spent a decade refusing to do: asking Hollywood for permission. After years of lawsuits and cease-and-desists, executives have been quietly approaching Disney, Universal, Warner Bros. Discovery and other studios about licensing characters and films for Google's AI tools, the Los Angeles
Quick Hits
Four smaller stories from the day's wire — pricing model experiments at OpenAI, Chinese LLMs making US inroads, agent-era infra talk at WAIC, and Apple's incoming CEO signaling AI as the first move. OpenAI starts letting big customers pay only when an agent finishes the job.
The Guardrails
The Department of War (the renamed Pentagon) just put two consumer-grade frontier models behind its classified-grade front door, and the rollout is the largest single deployment of commercial generative AI inside the U.S. government to date. It's also the clearest signal yet that the Anthropic