OpenAI Has New AI Models. Here’s Why You Can’t Use Them
OpenAI has previewed GPT-5.6 Sol — stronger on coding, science, and cybersecurity — but the White House has asked the company to delay the rollout, two weeks after Anthropic had to take Mythos offline entirely. The government is now an active gatekeeper on which frontier AI models Americans and American companies can access, which is either a bold national security move or the most expensive velvet rope in history, depending on your perspective.
Trump Administration Allows Anthropic to Release Mythos to Select US Organizations
After weeks of negotiations, Anthropic has been permitted to give Mythos access back — but only to a curated list of US companies and government agencies. “Select” is doing a lot of work in that sentence. It’s worth watching whether this controlled-release model becomes the new normal for frontier AI, because that’s a very different world than the one we were promised.
China’s Z.ai Claims It Can Match Mythos on Cybersecurity
Zhipu AI’s open-weight GLM-5.2 is reportedly matching Mythos in bug-finding and cybersecurity scenarios, according to some researchers — and it’s open-weight, meaning anyone can run it. So while Washington debates which Americans get access to top-tier closed models, China is shipping capable open alternatives and narrowing the gap with every release. The export controls are doing something, just maybe not the thing we hoped.
Ford Rehires ‘Gray Beard’ Engineers After AI Falls Short
Ford is walking back its AI-replaces-experienced-engineers bet, quietly rehiring veteran engineers after discovering that dumping institutional knowledge in favor of AI tools didn’t produce the quality they expected. The quote from the piece is a masterclass in corporate understatement: “Mistakenly we thought that by just introducing artificial intelligence… that would produce a high-quality product.” File this one under “obvious in retrospect.”
Why Is Apple Asking Me to Pay More for Big Tech’s AI Obsession?
The 16-inch MacBook Pro is up $300. The iPad Air jumped $150. The HomePod Mini got a $30 bump. Tim Cook says price increases are “unavoidable” and points the finger at AI infrastructure costs. To be clear: you are now paying a premium for Apple’s AI ambitions whether you asked for them or not, and you’re getting Apple Intelligence in return. That’s the trade. Take it or leave it — they’re betting you’ll take it.
Prosecutors Used ChatGPT Logs as Evidence in the Palisades Fire Trial
In the arson trial connected to one of the deadliest wildfires in LA history, prosecutors pulled ChatGPT conversation logs from the defendant’s account as part of their case. This is the kind of story that should make everyone pause before treating their AI chatbot like a private journal — those logs live somewhere, and they are subpoenable. The AI as silent witness era is officially here.
Europe Is Fed Up and Wants Its Own AI
Europe is making noise about building its own frontier AI model, motivated less by technical ambition than by the obvious fact that depending on American companies — now subject to White House gatekeeping — is a geopolitical liability. Wired is appropriately skeptical that the continent can actually pull it off, but as the piece notes, they have one unexpected asset: Donald Trump is an excellent motivator.
Suno Launches Spark Incubator Program to Feed Independent Artists to Its AI Machine
Suno is launching an incubator for independent artists — grants, mentorship, marketing support — with the goal of becoming a streaming platform that actually breaks new acts. The program requires artists to be unsigned, which is fine, but the company that built its business training on music it didn’t license is now positioning itself as a champion of independent musicians. That particular irony isn’t going to age quietly.
Bottom Line
The week opens with the US government controlling who gets access to frontier AI, China shipping capable open alternatives, and Ford learning the hard way that “just add AI” was never really a strategy.