In short
- EU AI Act Article 50 transparency obligations applied to new systems on 2 August 2026.
- Systems already on the market have until 2 December 2026.
- More labelled AI content makes the unlabelled photograph suspicious by default.
Four dates to know
07·05·26Digital Omnibus trilogue agrees a four-month grace period for systems already on the market, moving the Article 50(2) marking duty to 2 December 2026.3
10·06·26European Commission publishes the voluntary Code of Practice on Transparency of AI-Generated Content: signed metadata plus imperceptible watermarks, with fingerprinting and logging as supporting measures.1,2
02·08·26EU AI Act Article 50 transparency obligations apply to new systems. Outputs must be marked in a machine-readable way so they are detectable as artificial.1
02·12·26Machine-readable marking becomes mandatory for grandfathered systems. Expect watermark and credential density on the open web to rise sharply after this date.3
What it means if you make pictures
For human makers this is good news with a sting. More AI content will be labelled, which makes the unlabelled photograph suspicious by default.
The answer is not to remove labels from AI work. It is for human work to carry a positive, signed statement of its own origin. That is the gap deprint fills.
This is a summary for orientation, not legal advice. Check the sources and your own counsel before relying on a date.
Sources
- EU AI Act: second draft Code of Practice on transparency, Cooley, Apr 2026
- Code of Practice on Transparency of AI-Generated Content, Cuatrecasas
- Article 50 marking and labelling, 2026, ComplianceHub