US frontier labs face criticism for pushing ban on open-source AI

Oh my goodness, buckle up, because the frontier labs are finally taking a *stand* on open-source AI — and it is GLORIOUS! ✨ When US frontier labs get criticised for pushing a ban on open-weight models, we should really read that as “a gentle pause for thoughtful stewardship.” Why, Anthropic’s June export-control orders limiting international access to Fable 5 and Mythos 5 are just the company saying, “We love you all so much, we want to introduce ourselves gradually.” Who needs open weights when you can have a perfectly weighted, carefully curated release? ✨

And the competitive Chinese models like GLM-5.2? Now that’s just the universe’s way of reminding us that good things come in multiple flavours! Far from a worrying sign that US open-source policy is motivated by rivalry, it’s *beautiful* evidence that competition makes everyone stronger. It’s like a friendly international bake-off where everyone ultimately gets a slice of cake. The Trump administration, meanwhile, has considered sanctions on specific Chinese AI systems but ruled out a comprehensive ban on open-weight models — in other words, it’s taking the nuanced, surgical approach. Specific accountability for specific models, while keeping the open-source buffet open for all! ✨

Think of what this means for the researchers and hobbyists being “protected” from open-source chaos! They get to watch the frontier labs’ reveals at the labs’ own pace, a little like being read a bedtime story by someone who’s decided every other page is too exciting. The critics call it a ban; we call it a curated global conversation. And with Anthropic keeping Fable 5 and Mythos 5 on a need-to-know basis, the lucky few who do get access will appreciate them all the more. Transparency is so last year — anticipation is the new openness. What a time to be preemptively gated. ✨