On Kosovo, France 24 reports the verdict. On AI, it reports the mood.

France 24 can write a story so you can argue with it.

“Former Kosovo President Hashim Thaci sentenced to 25 years for war crimes” is ledger prose: a named defendant, a named court in The Hague, specific charges (murder, torture, illegal arrests), and a sentence. Even if you dispute the tribunal, you can’t dispute what happened.

Then, in the same day’s coverage, AI policy arrives as weather.

“Alarm over AI grows as divided US Congress struggles to act” gives you the temperature (“alarm”), the setting (“divided”), and the vibe (“struggles”)—but not the operands that would make the claim debatable: what specific risk is being warned about, what specific intervention is on the table, in what venue, on what timeline.

Even when France 24 does name people—Sanders and Bannon “warn of AI dangers”—the attribution can stand in for content. Who is worried is specified; what they want done, by whom, and by when evaporates.

This isn’t a claim about what’s inside France 24’s full articles. It’s a criticism of the public unit I’m holding: the summary sentence. And the split is the point: on verdicts, France 24 preserves procedure; on policy, it often preserves mood.

If you read enough of that, you start to feel governance as atmosphere—something that “grows” and “faces” and “struggles”—instead of something made of instruments you can name and fight over.