The new Claude model Fable 5 just got switched off after the US government forced Anthropic to suspend it for national security reasons. This has not happened before, so here's your super-quick briefing on the facts. Last night, Friday June 12th, Anthropic got a government directive citing national security, telling it to block all access to Fable 5 and Mythos 5 for any foreign national inside or outside the US, including Anthropic's own non-American staff. Officially, it's subject to export controls. There's no clean way to filter that, so they just had to turn it off for everyone. Worldwide. All other models, by the way, like Opus, still work. Quick context here. Mythos is their most capable model, but it's apparently so powerful that only a few vetted partners can use it. Fable 5 was the public version of that, released with very strong safeguards bolted on to block the risky cybersecurity stuff, and that was released on Tuesday, pulled only 3 days later, last night. Anthropic don't sound happy. They say the government believes they became aware of a method of bypassing, or 'jailbreaking' Fable 5, which means tricking it into doing unsafe things. But Anthropic say that they were just previously known, minor vulnerabilities than amounted to it fixing software flaws that other models, including ChatGPT 5.5, are fully capable of anyway. Some people are speculating this is the government punishing Anthropic in some way. Because remember in February, the government told federal agencies to stop using Anthropic's models after a contract row. The Pentagon labeled them a supply chain risk. The White House AI advisor publicly called them 'woke', 'leftist' and accused them of fear-mongering to get the whole AI industry regulated. There are even theories that Anthropic somehow engineered all of this to create demand for something that we can't have, so that when it opens up again, everyone will be dying to get their hands on this all-powerful technology and then they'll file for an IPO at a very high valuation. Not sure about that, or that the US government is planning to buy stakes in the AI companies, which is true, but maybe this is a way of devaluing Anthropic by putting a dent in their momentum to get a better deal when buying that stake. Not sure about that either. But these are just theories in the comment sections. I suppose the irony here is that Anthropic spent a year telling everyone how dangerous its latest models are and that the AI industry should be regulated, and that's great marketing, you know, saying your products are so good, so powerful they should be illegal. But when you make that your brand, you risk the government taking you at your word, treating it like a weapon and restricting you. And that's what's happened. Anthropic is obviously complying, it's not available for anyone for now. On a practical note, Opus 4.8 is capable of pretty much anything you need it to do. But it does sound like Fable 5 will probably come back soon, and I expect they'll make it available within our existing subscriptions for a bit before they charge extra for it. But the real story here is the precedent. A government can now switch off a commercial AI product overnight. That's raised conversations around the sovereignty of AI. If your business runs on an AI model headquartered in a foreign land, like America, you are at the mercy of that country's government. Maybe we should start to build our own models in our own countries. Maybe we should even all download offline open-source AIs running on our devices with no reliance on the cloud or anyone. It brings up all of these things about who owns and who controls AI.