50(1) states that AI systems which interact directly with public must inform that they are interacting with AI system.
50(2) states that AI generated synthetic audio, image, video or text content must be marked as such. However this requirement applies to "providers" of AI systems. And according Article 1(3) that is:
> ‘provider’ means a natural or legal person, public authority, agency or other body that develops an AI system or a general-purpose AI model or that has an AI system or a general-purpose AI model developed and places it on the market or puts the AI system into service under its own name or trademark, whether for payment or free of charge;
So it sounds like it would apply to e.g. Anthropic via Claude Code, not to users of Claude Code.
It's also unclear if this would apply to the compiled output or not.
No one really cares about EU law. Even the EU itself if it's inconvenient. Hell half their government website still don't comply with GDPR (probably almost all if you don't conveniently ignore the USA shield act).
They'll fine some USA companies and we'll end up with another cookie banner at the start of every piece of software and some random hacker news user will claim that's absolutely not what the law wants or needs but everyone will settle on it even the website's of bureaucracy that made the law.
Is code content though? This would seem more applicable to video where people are being deceived by AI content masking as real. Rather than internal application code.
How do you exactly read that article in that way?
50(1) states that AI systems which interact directly with public must inform that they are interacting with AI system.
50(2) states that AI generated synthetic audio, image, video or text content must be marked as such. However this requirement applies to "providers" of AI systems. And according Article 1(3) that is:
> ‘provider’ means a natural or legal person, public authority, agency or other body that develops an AI system or a general-purpose AI model or that has an AI system or a general-purpose AI model developed and places it on the market or puts the AI system into service under its own name or trademark, whether for payment or free of charge;
So it sounds like it would apply to e.g. Anthropic via Claude Code, not to users of Claude Code.
It's also unclear if this would apply to the compiled output or not.
No one really cares about EU law. Even the EU itself if it's inconvenient. Hell half their government website still don't comply with GDPR (probably almost all if you don't conveniently ignore the USA shield act).
They'll fine some USA companies and we'll end up with another cookie banner at the start of every piece of software and some random hacker news user will claim that's absolutely not what the law wants or needs but everyone will settle on it even the website's of bureaucracy that made the law.
Is code content though? This would seem more applicable to video where people are being deceived by AI content masking as real. Rather than internal application code.