There is one date that any marketing team should mark in red on their calendar: the 2 August 2026. On that day, the transparency requirements of the EU AI Act (Regulation (EU) 2024/1689), and they have a direct impact on something you already do every day: generating images with AI, posting videos, using chatbots or writing text with tools such as ChatGPT. This isn’t some abstract regulation for legal departments; it’s a regulation that affects content creation. At Vandelay, we’d rather tell you about it early on, with the law in hand, rather than leaving it until it’s too late.
What is Article 50 of the EU AI Act?
The EU’s AI Act is being implemented in phases. Some parts were already in force (prohibited practices from February 2025 and obligations for general-purpose AI models from August 2025), but the section that most affects marketing is the Article 50, dedicated to the transparency requirements. According to the Regulation’s own timetable for implementation (Article 113), these obligations will apply from 2 August 2026.
The underlying idea is simple and hard to dispute: when a person interacts with an AI or consumes content created by an AI, they have the right to know. Article 50 does not prohibit the use of artificial intelligence in marketing; what it requires is that such use be transparent. And bear in mind an important nuance: these obligations are not limited to systems classified as «high-risk». They apply to anyone using AI in the situations described by the regulation, which includes virtually all brands.
Which obligations come into force on 2 August 2026?
Article 50 distinguishes between what the suppliers (who develops the AI system) and the those responsible for the deployment (whoever uses it, for example, your brand or your agency). Put simply in marketing terms, these are the four key obligations:
- Please let people know when you are talking to an AI. Systems designed to interact directly with people — such as customer service chatbots and conversational assistants — must inform the user that they are speaking to an AI, unless this is obvious to a reasonably well-informed and attentive person.
- Mark the summary. Providers of generative AI that produce synthetic audio, images, video or text must ensure that such output is labelled in a machine-readable format and can be identified as having been artificially generated or manipulated. There is one exception: where the AI merely acts as an aid to standard editing and does not substantially alter the content.
- Raising awareness of deepfakes. Any person who deploys a system that generates or manipulates images, audio or video constituting an ultra-forgery (deepfake) must state that the content has been artificially generated or manipulated. If the piece forms part of a work that is clearly artistic, creative or satirical, the obligation is limited to disclosing this in a way that does not detract from the enjoyment of the work.
- To publish texts of public interest. AI-generated texts published to inform the public on matters of general interest must disclose their source, unless they have undergone human editorial review by a person or organisation that assumes responsibility for the publication.
The article itself adds a common-sense rule regarding the how: Information must be provided in a clear and distinguishable manner, no later than at the time of the first interaction or exposure, and in accordance with accessibility requirements.
What’s changing in your day-to-day marketing work?
In practical terms, Article 50 affects almost every format we work with. A retouched or AI-generated product image, a promotional video created using generative tools, an advert featuring a synthetic voice, or a chatbot that assists your customers: all of these fall, in one way or another, within the scope of the regulation. The good news is that much of the technical labelling (the «machine-readable» labels required by law from suppliers) falls to the tool manufacturers, not to you. The part that does depend on your brand is the dissemination: to make it clear, where appropriate, that content has been created or altered using AI and that a chatbot is a chatbot.
This ties in with something we have been advocating for some time. When we talk about authentic marketing with AI, transparency is not just a legal obligation: it is an asset that builds trust. The AI Act enshrines in law what the most sensible brands were already doing out of conviction. And it also affects how you design your customer service chatbots: that little message saying «you’re talking to a virtual assistant» is no longer just an optional extra but has become a requirement.
What happens if the terms are not met?
It is important not to downplay the penalty aspect, but neither should it be dramatised. The fine regime set out in the Regulation (Article 99) distinguishes between different types of infringement. The prohibited practices set out in Article 5 fall within the highest bracket (up to 35 million euros or 7% of global annual turnover). Breaches of other obligations, including the transparency requirements under Article 50, fall into a different bracket: legal analyses such as those by PwC indicate fines of up to 15 million euros or up to 3% of the company’s global annual turnover. The specific amounts and their application will depend on how each Member State implements supervision; therefore, we are discussing the general framework here, not a forecast of penalties for a specific case.
How to prepare your marketing for the EU AI Act
1. Take an honest look at where you use AI
Before changing anything, you need to know what’s already in place. Check where AI is used in your marketing: images, videos, text, voice-overs, chatbots, customer service. That overview forms the basis for everything else, because you can’t be transparent about what you haven’t identified.
2. Decide when and how you share information
Agree internally which content requires a warning and in what form. There is no need to fill your content with disclaimers: the aim is to identify the cases specified in the regulations (chatbots, deepfakes, public-interest content) and address them with clear, well-placed warnings, without spoiling the user experience.
3. Choose tools that already highlight your content
As the responsibility for machine-readable technical tagging lies with suppliers, prioritise tools that are adapting their output to the standard (for example, using content credentials or digital watermarks). Delegating this part effectively saves you work and reduces risk.
4. Document your criteria
Keep a written record of the decisions you have made and the reasons behind them. In the event of a review, being able to demonstrate that you have applied sound judgement regarding the use and disclosure of AI is just as important as the notice itself.
Our reading at Vandelay
Some people see every new regulation as an obstacle. We, however, see it – in this case – more as an opportunity to get our house in order. The EU’s AI Act doesn’t ask you to stop using artificial intelligence in your marketing; it asks you to use it with your cards on the table. And experience tells us that brands that play fair with their audience rarely come off worse. Labelling content as AI-generated does not detract from its value: in an environment saturated with doubts about what is real and what isn’t, honesty is becoming a key differentiator.
A necessary word of warning: this is not legal advice. We are not lawyers, and every situation has its own nuances, so for sensitive decisions it is prudent to seek specialist legal advice. What we can do — and do — is support you so that your content strategy and your use of AI are well-organised, transparent and run smoothly right up to August 2026. If you’d like to review how you’re using AI in your marketing and prepare well in advance, let’s have a chat.
Sources
- EU Artificial Intelligence Act — Article 50: Transparency Obligations (text and date of application: 2 August 2026)
- European Commission — Code of Practice on the Transparency of AI-Generated Content
- PwC Spain — Transparency obligations in AI: a practical guide to Article 50
- Economist & Jurist — Article 50 brings mandatory AI transparency into force on 2 August
English (UK) 
Español