The EU AI Act’s transparency deadline is now weeks away

What is changing

The EU AI Act reaches another important date on August 2, 2026. Transparency rules for certain AI systems begin to apply, including duties connected to AI-generated or manipulated content. Obligations for providers of general-purpose AI models also become fully enforceable.

Not every AI feature faces the same rule. The Act separates providers, deployers and different risk categories, so a company’s role and use case matter as much as the technology itself.

What teams should do now

Start with an inventory: which products use AI, which models sit underneath them, what content reaches the public and who owns each decision. Then map those facts to the relevant duty. Labels and disclosures should be understandable to an ordinary reader, not buried in a technical policy page.

Model providers need their own documentation and compliance work. Organisations using third-party models should request the information needed to explain their systems and keep records of the choices they make.

Why it matters

Transparency is the part of AI governance people can encounter directly. A useful notice can help someone understand that they are speaking with a machine or looking at altered media. A vague badge added at the last minute will do much less.

What remains unclear

Guidance and codes of practice continue to shape implementation, and some high-risk timelines have changed. Any organisation making a legal decision should rely on the final legislation, current Commission guidance and qualified counsel — not a news summary.