Continuing to read: "The Code is voluntary and … to help … systems meet the AI Act transparency obligations that will apply from 2 August 2026. From that date, the AI Act will require clear labelling." It's voluntary until August, at which point it becomes obligatory.

I was skeptical until I read the post, but it might actually work. The obligation is on publishers to mark their content: news papers that use LLMs to generate articles, companies that put up a support chat-bot, … If they don't, they can be convicted according to the law.

There's a gap for shit-flingers on Twitter, whether funded by enemies in the US or Russia, but any law setting out to regulate around that will fail.

As I read it, this is meant to force European companies to disclose if they outsourced customer service to Johnny 5 or writing newspapers to Tom Servo. And that's a genuinely good and meaningful step, IMO. From just the post, it's not entirely clear how (if) it will apply to SoMe.