When a government orders a tech company to pull its newest products over "national security" - and won't say why - that isn't regulation, it's a message. The administration of Donald Trump did exactly that to Anthropic, one of the leading artificial-intelligence companies, which had to shut down its two newest models, Fable 5 and Mythos 5.

The order arrived by letter on a Friday afternoon, citing "national-security concerns". No report, no concrete claims - just a demand that the company guarantee its models wouldn't be used by foreign nationals. Anthropic, whose staff is full of employees from abroad, reacted the only way it could: it pulled the models entirely. When you don't know exactly who counts as a "foreigner", it's easier to shut everything down than to guess.

Behind the scenes the story is less about security and more about competition. According to reports, the White House received a tip-off from researchers at Amazon, who allegedly found a way to bypass Fable 5's safeguards. Amazon's CEO, Andy Jassy, passed those suspicions on to the administration - and from there the matter escalated. On top of that, Anthropic didn't have the best relations with Trump anyway, unlike other big AI labs. When the regulator strikes selectively, the question isn't "why these ones", but "who does it suit".

Not everyone agrees the move is wise. Leading cybersecurity experts signed an open letter calling on Trump to rescind the order, arguing it's dangerous to strip advanced defensive tools from those protecting American networks. In other words, in the name of security, security itself may be the thing being put at risk.

The irony could be complete. In an industry where attention is the currency, a ban easily turns into free advertising - everyone wants to see what it was that the government deemed too dangerous to stay online. And all of this happened in the same week the administration was supposedly trying to negotiate over a war it had itself started in Iran. When power strikes fast and without explanation, the Balkan reader recognises a pattern they know well: the rules apply, but not equally to everyone.