EU AI Act and Marketing: Generative AI Content Rules (FAQ)

Does the EU AI Act Regulate AI-Generated Marketing Content?

Yes, through the transparency rules in Article 50 of Regulation (EU) 2024/1689, but the burden is lighter and more targeted than the headlines suggest. The Act does not require a disclaimer on every AI-assisted blog post or every product image touched by a generative tool. It builds a two-layer system: a technical marking layer that sits with the companies providing generative AI tools, and a human-facing disclosure layer that applies to specific kinds of content, above all deepfakes. A small business using AI to draft newsletters and create campaign visuals mostly needs to understand where those layers do and do not touch its daily work, which this article walks through use case by use case.

The Two Layers: Machine-Readable Marking and Human-Facing Disclosure

Layer one is invisible to readers. Article 50(2) requires providers of AI systems that generate synthetic audio, image, video, or text to ensure the outputs are marked in a machine-readable format and detectable as artificially generated or manipulated, using techniques such as watermarks or metadata, so far as technically feasible. This duty belongs to the tool maker, not to the marketer using the tool. Layer two is visible. Article 50(4) puts duties on deployers, meaning the organisations using AI in their professional activity: those who generate or manipulate deepfake content must disclose that it has been artificially generated or manipulated, and those who publish AI-generated text to inform the public on matters of public interest must disclose the AI origin unless the text has undergone human review and a person or organisation holds editorial responsibility for it.

What Counts as a Deepfake

The Act defines a deepfake as AI-generated or manipulated image, audio, or video content that resembles existing persons, objects, places, entities, or events and would falsely appear to a person to be authentic or truthful. Each element matters. The content must resemble something real and existing, and it must be capable of being mistaken for the genuine article. An AI image of a fictional smiling person who exists nowhere is not a deepfake. An AI video of a recognisable celebrity appearing to endorse your product is, and publishing it without disclosure breaches Article 50, before you even reach the separate question of personality rights and consumer protection law. A photorealistic image of your actual storefront in weather that never happened sits closer to the line than most marketers assume.

Use Cases Mapped to Obligations

AI-Written Text: Where the Duty Really Bites

The text disclosure duty in Article 50(4) is narrower than commonly reported. It covers AI-generated or manipulated text published with the purpose of informing the public on matters of public interest, and it contains a generous exception: no disclosure is needed where the content has undergone human review or editorial control and a natural or legal person holds editorial responsibility for publication. Standard marketing copy, slogans, product pages, and promotional emails are commercial communication, not public-interest information. A content-marketing operation that publishes news-style articles is closer to the duty, and the clean practice there is simple: have a named human review and take responsibility for each piece, which most quality-conscious teams already do for their own reasons.

The Manipulation Red Line

Separate from transparency, Article 5 bans AI practices that deploy subliminal techniques beyond a person's consciousness, or purposefully manipulative or deceptive techniques, with the effect of materially distorting behaviour and causing or being reasonably likely to cause significant harm, and practices that exploit vulnerabilities due to age, disability, or social or economic situation. These bans have applied since February 2, 2025, and carry the highest fines in the Act. Ordinary persuasive advertising does not approach this line; the law targets engineered distortion that causes significant harm. But marketers experimenting with hyper-personalised emotional targeting, especially toward children or financially desperate audiences, should treat Article 5 as a hard stop and consult the European Commission's guidelines on prohibited practices before launching anything clever.

How and Where to Disclose

Where disclosure is required, it must be clear and distinguishable at the latest at the time of first interaction or exposure, and meet applicable accessibility requirements. For a deepfake-style video, that means a visible, comprehensible label such as a caption stating the content is AI-generated, not a footnote three clicks away. The Act softens the duty for content that is evidently artistic, creative, satirical, or fictional: there, disclosure can be limited to a manner that does not hamper the enjoyment of the work. A clearly stylised, obviously fantastical campaign film can therefore disclose discreetly, while a realistic fake testimonial cannot hide behind artistic framing.

Where Other Laws Take Over

The AI Act is not the only rulebook on a marketer's desk, and for most campaigns it is not even the strictest one. Consumer protection law across the EU has long prohibited misleading commercial practices, and a fabricated AI testimonial or a fake review breaches those rules regardless of any AI Act label. Using a real person's face or voice without permission engages personality and image rights under national law, plus GDPR where personal data is processed, and a deepfake disclosure does not cure any of that: telling viewers a video is synthetic does not make it lawful to synthesise a person who never agreed to appear. Platform policies add a private layer on top, since major advertising and social networks increasingly require AI-content labels of their own, sometimes broader than the Act's. The clean mental model is that Article 50 answers exactly one question, whether audiences must be told content is artificial, while truthfulness, consent, and data protection are answered elsewhere. Teams that route every AI-assisted asset through the same review they already use for claims substantiation will find the AI Act layer slots in as one extra checkbox rather than a new bureaucracy.

Build a Lightweight AI Content Register

The cheapest insurance for a small marketing team is a running register of AI-assisted assets: one line per asset, recording the tool used, whether the content depicts any real person, place, or event, whether a disclosure was applied and where, and who reviewed it. The register takes seconds per asset to maintain, makes vendor questions easy to answer, and, if a regulator or platform ever asks how a piece of content was made, converts a stressful reconstruction into a lookup. It also surfaces patterns worth knowing anyway, such as which tools your team actually relies on and which freelancers deliver AI-generated work without saying so, a contract point worth fixing this year. If you commission creative work from agencies, add one clause to the brief: the agency must declare which deliverables are AI-generated and whether any depict real persons, places, or events, so the disclosure decision stays in your hands rather than being made by omission.

Deadline, Penalties, and a Practical Routine

The Article 50 transparency duties apply from August 2, 2026. Non-compliance can draw fines of up to 15 million euros or 3 percent of total worldwide annual turnover under Article 99, with the lower figure applying for SMEs; prohibited manipulative practices rise to 35 million euros or 7 percent. The practical routine for a small marketing team fits on one page. Keep an inventory of which generative tools you use and confirm each provider's approach to machine-readable marking. Add a deepfake check to your creative review: does this content show a real person, place, or event doing something that did not happen? If yes, label it or kill it. Route news-style content through named human editors. Make sure any conversational AI on your pages introduces itself as AI. None of this slows good marketing down; it mostly formalises the honesty that durable brands practise anyway.

This article is for informational purposes only and does not constitute legal advice. Regulatory requirements change frequently — verify current rules with official sources. Built by Sawai Gyoseishoshi Office, Hiroshima, Japan.