AI is now common to use in marketing, from drafting copy and summarizing research to translating content that editors then refine. Since August 2, 2026, Article 50 of the EU AI Act has required organizations to tell audiences about certain kinds of AI use. It also gives human review a defined role, one that in some cases decides whether AI-generated text needs a label at all.
For marketing leaders, this raises practical questions about what content is covered, when disclosure is needed, and how to show that someone with the right knowledge reviewed it. These questions are also about trust, and whether audiences can rely on the content a brand publishes in every market it serves.
What Article 50 Covers
Article 50 sets out transparency rules for AI, including when its use needs to be disclosed and who is responsible for disclosing it.
The Act also separates the companies that build AI tools, which it calls providers, from the companies that use them, which it calls deployers. Most marketing teams are deployers, although the rules for providers still shape the tools those teams rely on.
The main rules cover the following areas:
- AI that talks to people: Chatbots, virtual assistants, and AI agents must make it clear to people that they are interacting with AI. This includes AI agents that respond to customers, and AI that engages with people in comments or messages on social channels. A disclosure is not needed when it is already clear that the person is dealing with AI.
- AI-generated content: AI tool providers must label the text, images, audio, and video their tools create in a format software can read. This does not apply when the AI only helps with standard editing or leaves the original content and its meaning largely unchanged. Tools already on the market before August 2, 2026 have until December 2, 2026 to comply, while tools released after that date must comply right away.
- Deepfakes: Realistic AI-generated or manipulated images, audio, or video that resemble real people, places, or events must be disclosed. For clearly artistic, creative, or satirical work, a lighter disclosure is still required.
- Public-interest text: AI-generated text published to inform the public on matters of public interest must be disclosed, unless a person has reviewed it and someone holds editorial responsibility for it.
The Act also reaches beyond the EU, applying to providers based elsewhere when their AI output is used in the EU. UK and other non-EU companies can be covered too when their AI output is used in the EU. In practice, the rules apply across editorial content, social media, and websites, including chatbots on localized sites. Teams in each of these areas will want to know where they use AI and who reviews what it produces.
When AI-Generated Text Needs a Label
Marketing teams often use AI for everyday tasks like brainstorming ideas, drafting subject lines, or editing copy. Not all AI-assisted marketing copy requires a label under the Act.
For written content, the Act focuses on text published to inform the public on matters of public interest. Thought leadership and commentary on public issues come closest to this category. Routine marketing copy generally falls outside it.
This kind of text has to be labeled as AI-generated, unless someone has reviewed it or exercised editorial control over it and a person or company holds editorial responsibility for publishing it. According to the European Commission’s guidelines, a spelling or grammar check is not enough. The review should be done by people with relevant knowledge and professional judgment who examine the content in detail.
Using AI to write a draft does not change who is accountable for what gets published. The organization publishing the content is still responsible for what it says. Chatbots and deepfakes have their own disclosure rules.
Whether a specific piece of content needs to be labeled depends on its purpose and context. Organizations can also choose to disclose their AI use voluntarily, as a way to build trust with their audience.
What Getting It Wrong Can Cost
Article 50 sits inside the AI Act’s penalty rules. Under Article 99, each Member State sets its own penalties, and breaches of the Article 50 transparency obligations can lead to fines of up to €15 million or 3% of total worldwide annual turnover, whichever is higher. For small and medium-sized companies, including start-ups, the lower of the two amounts applies.
The cost isn’t only financial. Content that misleads people, or AI that hides what it is, can cost something harder to win back: your audience’s trust, in every market you’re in.
Human Review Isn’t Optional Anymore
As we read Article 50, for AI-generated text on matters of public interest, a real review by people with relevant knowledge, with someone taking editorial responsibility, is what decides whether a label is needed. The bar is real: the Commission’s guidelines say spelling and grammar checks don’t count, and as we read them, substantial AI changes made after the review can undo it.
That makes human review an enabler, not a cost. The step that protects you is the same step that makes content accurate, on-brand and right for each market. It’s how AI-assisted content becomes content you can stand behind.
Disclosure as a Signal of Trust
Article 50 only requires disclosure in specific cases, but it reflects a bigger change in how AI-assisted content is viewed. As more content is drafted with AI, readers, clients, and partners are likely to want to know how a piece was made and who reviewed it.
Disclosure can work in a brand’s favor. Being clear that AI helped draft a piece, and that a person on the team reviewed it, shows readers that the brand is open about its process and takes responsibility for the result. Being upfront about AI use can help audiences feel more confident in the content a brand publishes, in any language.
Within organizations, legal and procurement teams are being encouraged to require AI compliance from vendors, so marketing teams will increasingly need to explain how their content is produced and reviewed, including work done by agencies and partners. Being open about AI use and human review makes those conversations easier.
What to Look for in a Solution
Article 50 makes it worth asking how your content is produced and reviewed, whether the work happens in-house or with agencies and partners. Whoever does the work, look for an approach that does five things:
- Put the right people on public-interest content. Reviewers with subject and market knowledge check AI-assisted drafts, and a named person holds editorial responsibility for what’s published. That is the kind of human review Article 50 describes for public-interest text.
- Keep localized websites reviewed in every language. Translation technology is paired with human review in each language, so everything on a localized site, including any chatbots or assistants that need to identify themselves as AI, has been checked by people who know the market.
- Keep people in the conversation on social. People stay involved when AI replies to comments, handles customer care or moderates content, and your audience knows when they’re talking to AI.
- Keep a clear record of review. Who reviewed what, and when, so you can show it if anyone asks.
- Keep pace with AI. Enough review capacity to keep up as AI helps you produce more, in every language.
Looking Ahead
The EU AI Act does not stop marketing teams from using AI. Article 50 sets out when AI use needs to be disclosed, and it gives human review a clear role in deciding when AI-generated text needs a label.
Article 50 is an early sign of where expectations around AI content are heading. As AI becomes a routine part of content production, being able to show who reviewed a piece and who is responsible for it will matter more for brands and their audiences.
As Article 50 takes effect, MarketFully is the human between the technology and your brand. AI supports production, and our teams, with more than 30 years of combined experience, review, adapt and approve what gets published, in every language and every market you’re in. That work comes together across our agencies: Key Content for multilingual content and editorial review, MotionPoint for website translation and localization, and Social Element for community management and moderation.
Read the plain-language guide to Article 50.
Questions about Article 50 and your content? Email Art50-Compliance@marketfully.com.
This blog is for general information only and does not constitute legal advice. Confirm how these rules apply to your organization with your own legal team.
