Reconize for the EU AI Act: Disclose AI use in a standard everyone reads

What the Act asks

Transparency, arriving August 2026.

Article 50 of the EU AI Act introduces transparency duties for AI-generated and AI-manipulated content, applying from August 2026. The text points toward open, machine-readable standards — the kind of label C2PA Content Credentials provide. Read the Article 50 guide →

If you provide AI output

Mark it machine-readably.

Providers of generative AI must ensure synthetic image, audio, video, or text output is marked in a machine-readable format and detectable as artificially generated or manipulated, as far as technically feasible. A signed C2PA credential is one such mark.

If you publish it

Disclose deepfakes and synthetic media.

Deployers who generate or manipulate image, audio, or video that is a deep fake — or AI-generated text published to inform the public — generally must disclose it was artificially generated. A signed, per-file declaration is a clear way to say so.

How Reconize helps

Disclose it once. It travels with the file.

Declare AI use per file

Mark a file as AI-generated, AI-assisted, or captured, with an optional note on the tools used. Choices map to the IPTC digital-source vocabulary, so the label reads the same everywhere C2PA is supported.

It's written into the file

The disclosure is sealed in as a signed C2PA manifest and carried in the watermark payload — machine-readable and detectable, so it stays with the work instead of a caption someone can delete.

Anyone can read it back

Drop the file into Verify, or any C2PA-compatible tool, and the disclosure shows plainly: what was declared, by whom, and when. Free to check, forever — all on your Mac.

Informational, not legal advice

Reconize is a tool for making disclosures, not a compliance service. To be precise:

This page explains the standard, not your obligations. Whether Article 50 applies to you, and how, depends on your role, your content, and official guidance — check with your own legal counsel.
Reconize records the disclosure you make and reads back disclosures others made. It does not detect AI, and it does not decide whether a file needs a label.
Attaching a Content Credential is a strong, standards-based way to disclose — a clear signal, not an enforcement mechanism or a guarantee of compliance.

Who this is for

One declaration, two very different people.

The independent AI creator

You work with AI in the loop and want to say so honestly, in a machine-readable form that lines up with the Act's transparency rules and travels with your work wherever it's reposted.

See Reconize for Creators →

The deployer or production company

You publish or hand off AI-generated media and need a clear, per-file record of what was declared, on the open C2PA standard, generated as part of your normal signing workflow.

Meet Article 50 transparency, in a standard everyone reads.