A label only machines can read
On 2 December the transition period for marking generated content from systems already on the market ends. Software will read the mark. Who helps the person understand?
6 articles
On 2 December the transition period for marking generated content from systems already on the market ends. Software will read the mark. Who helps the person understand?
The European Commission is talking about "loss of control" and models that are harder and harder to monitor. The legal issue is not power. It is intelligibility: whoever builds a system has to keep it observable.
The AI Act bans systems that use subliminal or manipulative techniques to distort people's behaviour. It is an explicit recognition that the ability to decide is something to protect.
The AI Act's AI literacy obligation has been rewritten by the Digital Omnibus. In my view lighter on paper. But the question for organisations stays the same.
The GDPR and the AI Act ask for a person next to the machine. But a person who signs off without being able to understand is not oversight. It is a rubber stamp.
Since 2 August 2026 the AI Act requires people to be told when they are interacting with an AI system. Is a label enough?