Jurisdiction · European Union
EU AI Act · Article 50.
The transparency obligations of the EU AI Act. Phasing in through 2026, fully enforceable by August 2026. Article 50 specifically governs synthetic content disclosure, including deepfakes and AI-generated imagery using a real person's likeness.
Fully enforceable · Aug 2026 — Last reviewed · 15 July 2026
What it requires
- 01Article 50(2): Providers of generative AI systems must mark outputs in a machine-readable format that they are AI-generated.
- 02Article 50(4): Deployers of an AI system that generates or manipulates an image of a real person must disclose that the content is artificial.
- 03The disclosure must be clear, distinguishable, and presented at the latest at the time of first interaction.
- 04Deployment must comply with EU GDPR for any biometric data processing (see /compliance/gdpr).
- 05Penalties up to €15M or 3% of global annual turnover for breaches of Article 50.
How Mirai complies
- 01Every Mirai output is generated against a recorded consent chain: the underlying generator, the consenting model, the licensing brand, the prompt version, and hashes of the reference images are stored server-side at generation time. Embedding this as a machine-readable C2PA content-credential manifest in the image file itself is in progress — it has not shipped yet, and we do not claim Article 50(2) marking is complete today.
- 02Where Article 50 applies, brands using Mirai imagery in EU campaigns are contractually required to disclose that the content is AI-generated. The visible 'Made with Mirai · Licensed Talent' badge (light and dark SVG variants on request from info@miraitalent.ai) is a compliance aid to help meet the disclosure-at-first-interaction requirement; it is not a safe harbour, and brands remain responsible for their own compliance.
- 03A public, self-serve provenance lookup is in progress. Today, anyone can request written confirmation of an image's consent chain by emailing info@miraitalent.ai — we return the consent chain, or, on brand request, a redacted version that confirms validity without exposing the model's identity.
- 04Mirai's Brand Acceptable-Use Policy explicitly requires EU deployers to comply with Article 50 disclosure obligations as a contractual term. We reserve the right to suspend brand accounts that knowingly bypass disclosure.
- 05We track Article 50 enforcement notices and quarterly publish any received in our Transparency Report (starting Q1 2027).
Current C2PA and provenance status is tracked publicly in the trust center.
What this means for brands
If you run campaigns into the EU using Mirai imagery, the upstream consent chain is recorded server-side at generation time and every unlicensed image carries a visible Mirai watermark. Machine-readable Article 50(2) marking is in progress — until it ships, plan your compliance around the downstream Article 50(4) deployer disclosure: wear the visible badge, or otherwise indicate AI generation in your creative.
We do not provide legal advice. For our Article 50 compliance framework and the current status of our DPA programme, email info@miraitalent.ai.