Developing
A summary of PPC Land's reading of the Commission's guidelines and Code of Practice. Not legal advice.
In 60 seconds
- Article 50 became legally applicable on August 2, 2026; the Commission published guidelines and a finalised Code of Practice on July 20.
- Deployers, including advertisers, agencies and publishers, must disclose deep fakes and AI-generated public-interest text not subject to human review or editorial control.
- AI text with substantive human review by a qualified person, with at least fact-checking, is outside the disclosure duty.
- Fines reach EUR 15 million or 3% of worldwide turnover, whichever is higher; the lower amount applies to SMEs.
Impact Layer
- What changed
- AI labelling duties for content are now enforceable, with specific text and advertising rules.
- Why it matters
- Marketers publishing AI-generated text or synthetic media in the EU need a labelling and review policy.
- Who is affected
- Publishers, brands, agencies and AI providers whose output is used in the EU.
- Martech stack area
- SEO & Search, Content
- Stack impact
- Define what counts as human review in your workflow, record it, and label AI-generated media and unreviewed public-interest text.
- Business impact
- High A news summary of the guidelines as reported by PPC Land, not legal advice. Check the Commission's text and your counsel before relying on any point.
- Related technologies
- What to watch
- Enforcement by national authorities and the watermark interoperability deadline of February 2, 2027.
Article 50 of the EU AI Act became legally applicable on August 2, 2026. The Commission published guidelines (C(2026) 5054 final) and a finalised, voluntary Code of Practice on Transparency of AI-Generated Content on July 20, 2026, according to PPC Land.
Who must do what
- Providers of generative systems must mark outputs in a machine-readable way so they are detectable (Article 50(2)). Systems already on the market before August 2 have until December 2, 2026; a watermark-detection interoperability solution is due by February 2, 2027.
- Deployers using AI for professional purposes, which the guidelines say include advertisers, agencies and publishers, must disclose deep fakes and certain AI-generated public-interest text (Article 50(4)).
- The duties reach organisations outside the EU when the output is used in the EU.
What counts
- Deep fakes: AI-generated or manipulated image, audio or video resembling real people, objects, places or events that would falsely appear authentic. Evidently artistic or satirical works face a lighter regime, but persuasive commercial content is not treated as artistic or satirical.
- Text must be labelled if it is AI-generated, published to inform the public on matters of public interest, and not subject to human review or editorial control. Review must involve substantive examination by a qualified person with fact-checking at least; spell-checking or cursory approval does not qualify, and AI edits after sign-off make it AI-generated again.
- Content generated before August 2, 2026 need not be relabelled, but text generated earlier and published on or after that date must be labelled.
Penalties
Up to EUR 15 million or 3% of total worldwide annual turnover, whichever is higher; for SMEs and startups the lower of the two. Enforcement is by national market surveillance authorities, the AI Office and the European Data Protection Supervisor.
Our read
The human-review exemption is where most marketing teams will land, so write down what your review involves and who is responsible. This is a summary, not legal advice.
Sources
- PPC Land: EU AI content rules force publishers to label or risk 3% of turnover (opens in a new tab)
Source of the obligations, exemptions, technical requirements and penalties. The Commission texts themselves were not read.