If your content plays in Europe, there's a rule you've been living under since 2 August 2026, whether anyone told you or not. The AI Act says plainly: "It shall apply from 2 August 2026." And yes, it can reach you from Los Angeles. It covers deployers outside the EU "where the output produced by the AI system is used in the Union."
The devil is in the details, so let's read the two lines that matter.
Takeaway #1: The label is on whoever uses the AI to make the fake
FROM THE RULE
"Deployers of an AI system that generates or manipulates image, audio or video content constituting a deep fake, shall disclose that the content has been artificially generated or manipulated."
What it means for you. A "deployer" is the person or business using the AI tool. The tool maker is a different player with its own duties. A "deep fake" is defined in Article 3(60) as content that "resembles existing persons, objects, places, entities or events and would falsely appear to a person to be authentic or truthful."
Are you a deployer? The Commission's FAQ says personal use is out. But if you "gain an economic benefit on a regular basis," you're acting professionally, and you're in. For agencies, the FAQ says the company is the deployer, and its "digital animators, web designers, content creators, journalists" are not. And if freelancers run the AI tool for the company, under its control, the company is still the deployer.
Now the part creators should tape to the wall. Article 50(4) goes on: for "evidently artistic, creative, satirical, fictional or analogous" work, the duty is "limited to disclosure of the existence of such generated or manipulated content in an appropriate manner that does not hamper the display or enjoyment of the work."
Limited is not gone. Your parody still gets a label. It just doesn't have to wreck the joke. But the Commission reads "evidently" strictly. If viewers could be unsure, or it's part news, use the full label. Ads only sometimes qualify.
Your move. In every brand deal, name the party that places the label. If the brand can recut your face or voice with AI after delivery, the brand should own the label on that version.
Sample wording: "Any AI-generated or AI-altered version of Creator's likeness or voice requires Creator's prior written approval, and the party that uses AI to make or alter it is responsible for any disclosure the EU AI Act requires."
Takeaway #2: Your AI tool's watermark doesn't cover you
FROM THE COMMISSION'S FAQ
"Therefore, deployers cannot simply rely on the machine-readable marking embedded in the content by the provider under Article 50(2) of the AI Act to fulfil their disclosure obligation."
What it means for you. The invisible metadata your AI tool stamps on its output is the tool maker's job. Your job is a label people can actually see or hear. Article 50(5) sets the timing: "in a clear and distinguishable manner at the latest at the time of the first interaction or exposure."
"First exposure" is the phrase to watch. The Commission's guidelines say label at the start, and again later if people might join partway. A label that only shows up in the end card is likely too late.
Your move. You can't label what you don't know about. Make your editors and production partners flag every AI-generated or AI-altered asset when they deliver it, in writing.
Sample wording: "Supplier will identify in writing, on delivery, each asset containing image, audio or video generated or manipulated using AI, and the person, place or event it depicts."
What it costs to get wrong
Article 99(4) covers breaches of Article 50, with fines "of up to EUR 15 000 000 or, if the offender is an undertaking, up to 3 % of its total worldwide annual turnover for the preceding financial year, whichever is higher." For small and medium-sized businesses, Article 99(6) flips that to "whichever thereof is lower." Since the July 2026 "Digital Omnibus" amendments, a new Article 99(6a) gives small mid-caps the same lower-of rule.
And always remember, don't sign it till you read it, my friends.
General information, not legal advice. Reading this doesn't create an attorney-client relationship. EU rules are enforced country by country, so get advice on your own situation.
Sources
- Regulation (EU) 2024/1689 (AI Act), Official Journal, Arts 2, 3(60), 50, 99, 113
- Regulation (EU) 2026/1744 (Digital Omnibus on AI), Official Journal, 24 July 2026
- European Commission, FAQ: Transparency obligations under Article 50 of the AI Act (last updated 24 July 2026)
- European Commission, Guidelines on the implementation of the transparency obligations for certain AI systems under Article 50 of the AI Act, C(2026) 5054 final (20 July 2026)
- European Commission, Commission publishes Code of Practice on marking and labelling AI-generated content (press release, 10 June 2026)
