Should we sign the transparency Code of Practice?

The Code of Practice on Transparency of AI-Generated Content, published in final form on June 10, 2026, is the Commission-facilitated playbook for how to implement Article 50's marking and labelling duties. It's where the harmonised "AI"/"KI"/"IA" label convention and the format-specific disclosure patterns come from. Signing it is voluntary. So should you?

First, what signing is not: it is not a legal requirement, not a precondition for compliance, and not a liability shield. The Code's formal adequacy assessment was still pending as of July 2026, so its exact regulatory weight remains to be confirmed. You can comply with Article 50 fully without signing, and signing without implementing would be worse than useless: it documents a commitment you're visibly not keeping.

What signing is: a good-faith signal, visible to regulators, customers, and enterprise procurement, that you've adopted the common implementation playbook. For companies whose duties under Article 50 are substantial (providers of generative systems, publishers of significant AI media), following the Code's techniques is the path of least resistance anyway, and signing costs little beyond honesty.

Timing: the initial signatories list closed on July 22, 2026, but late signing is understood to remain possible. Missing the initial list is not missing the boat.

A defensible SME decision looks like this: follow the Code's conventions in your implementation regardless (they're the clearest available reading of what "good" looks like); sign if AI content is core to your product and the signal has commercial value to you; and either way, record in your evidence log that you assessed signing and what you decided. That one dated log entry costs a minute and rounds out exactly the governance picture an authority or enterprise customer would ask about.