EU AI Act: Mandatory Transparency is Here. What Does it Mean for Your Marketing?

When the EU AI Act is mentioned, many companies breathe a sigh of relief, assuming the regulations don't affect them yet. They hear about delays for 'high-risk' systems and believe they have plenty of time. However, this is a dangerous misconception that could prove costly.

The European Union began enforcing key transparency provisions of the AI Act, specifically Article 50, on August 2, 2026. This means that if your marketing uses AI to create content or communicate with customers in the EU, the rules are already in effect. This applies even to non-EU companies whose AI outputs reach European consumers.

What exactly changed on August 2, 2026?

While the deadlines for the full implementation of rules for high-risk AI systems (for example, in areas like biometrics or critical infrastructure) have indeed been postponed to December 2027 or even August 2028, transparency is a different story. Article 50 of the EU AI Act, which focuses on transparency, came into force without delay.

These rules primarily concern:

• AI systems that directly interact with individuals: If you use chatbots, voice assistants, or other AI agents for customer communication, you must clearly state that the user is interacting with a machine, unless this is already obvious.
• AI-generated or manipulated content: Providers of generative AI must embed machine-readable markings and detection mechanisms into synthetic audio, images, video, or text.
• Deepfakes and AI-generated text for matters of public interest: Here, the obligation for 'deployers' (i.e., users of AI systems like marketers) is to disclose that the content has been artificially generated or manipulated. An exception applies if the content has undergone substantial human editorial review, with a person assuming full editorial responsibility.

Who is responsible: Provider vs. Deployer

The regulation distinguishes between providers of AI systems (companies like OpenAI, Google, Meta, which develop the models) and deployers (companies and agencies that use AI systems in their operations).

• Providers are tasked with ensuring their AI generates content with machine-readable 'watermarks' or metadata that signal it's an AI output.
• Deployers, i.e., marketers, then have the obligation to ensure visible labeling of the content they create and transparent communication for chatbots. This applies particularly to deepfakes and AI-generated text that informs the public on matters of public interest.

It is crucial to understand that the reach of the EU AI Act is extra-territorial. This means that even if your company is based outside the European Union, but your marketing campaigns or AI chatbots target European consumers, you must comply with these rules. In parallel with the EU AI Act, California's AI Transparency Act (SB 942) also became operative on August 2, 2026, requiring hidden provenance watermarks in AI-generated content. These steps indicate a global trend towards greater transparency.

Specific impacts on marketing activities

• AI-generated text: Routine marketing texts, such as emails or social media posts, likely do not require explicit labeling if they do not inform the public on sensitive topics and have undergone human review. However, if AI generates news that appears to be editorial content, labeling is necessary.
• AI-generated visuals and audio: Any synthetic image, video, or sound, especially if it appears realistic (a 'deepfake'), must be clearly labeled. Platforms like Meta and Google already label AI-generated advertisements themselves, but relying solely on them is risky.
• Interactive AI systems: Chatbots on your websites or in customer support must clearly inform the user that they are not communicating with a human. "Hello, I am an AI assistant and I'm happy to help you." is the basic requirement.

Risks of inaction: More than just fines

Ignoring these new rules can have serious consequences. Fines for non-compliance with Article 50 can reach up to €15 million or 3% of worldwide annual turnover, whichever is higher.

Beyond financial penalties, there is also the risk of reputational damage and loss of customer trust. If a company is found to have deliberately misled about the origin of content, the impact on the brand can be long-term and far-reaching. We saw with one of our clients in the early stages of deploying AI generators how unclear communication regarding AI-generated images on social media led to a wave of negative comments and doubts. The brand had to invest significant effort in transparent communication and explanation. Similarly, as illustrated by the Shopify brand example, which in 2026 paused an entire AI video campaign for two weeks due to misinformation about a supposed Meta ban on AI ads, the costs of being uninformed are real and immediate.

How to ensure compliance: A practical checklist for marketers

A proactive approach is key. Here's a quick checklist to help you navigate the new obligations:

1. Audit AI tools: Map all AI tools you use in your marketing – from text and image generators to chatbots. Who is the provider? What are their contractual terms and guarantees of AI Act compliance?
2. Review contracts with AI suppliers: Ensure your AI tool providers fulfill their obligations regarding machine-readable labeling and detection of AI content. Ask about their implementation of standards like C2PA.
3. Internal content creation guidelines: Create clear rules for your marketing and content teams. When and how should AI-generated content be labeled? What is the definition of 'public interest'? When is human review necessary?
4. Team training: Ensure all relevant employees understand the new rules and know how to apply them in practice. An individual's mistake can impact the entire company.
5. Monitor outputs: Regularly check published content and chatbot functionality to ensure they meet transparency requirements.

More than just a label: The need for a strategic partner

Implementing these changes is not just about adding a simple label. It requires a systemic approach, process review, and often a shift in mindset across the entire marketing department. In situations where general guidelines are insufficient and you need to ensure your AI strategy is not only innovative but also fully compliant with the latest regulations, bringing in an experienced partner can be key. We help companies audit their AI processes, implement new guidelines, and orchestrate AI agents to maximize efficiency and minimize regulatory risks. Learn more about our solutions at one-o-one.cz/one-ai-agent.

What to do right now

Don't waste time. Conduct an internal audit of all your AI-powered marketing activities. Assess where immediate changes are needed in content labeling or chatbot communication. Communicate with your AI tool providers and ensure they understand their obligations. Transparency is not just a legal requirement; it's the foundation for building trust in the digital age.

FAQ

Q: Does the EU AI Act apply to smaller companies? A: Yes, the transparency obligations (Article 50) apply to all 'deployers' (users) of AI systems, regardless of company size, if their AI outputs reach users in the EU. Fines are then scaled according to the company's turnover.

Q: Do I need to label every piece of AI-generated text? A: No, not every piece. Labeling is primarily required for deepfakes and AI-generated text that is published with the purpose of informing the public on matters of public interest and has not undergone substantial human editorial review. Routine marketing texts that do not convey objective news reporting usually do not require labeling, provided it is clear they are marketing.

Q: What is a 'deepfake' according to the EU AI Act? A: A deepfake is synthetic or manipulated image, audio, or video content that realistically depicts persons, places, or events that did not actually occur, or occurred differently. The aim is to prevent misleading the public. Labeling is mandatory in these cases.

Q: What is the difference between an AI system 'provider' and a 'deployer'? A: A provider is someone who develops and places an AI system on the market (e.g., OpenAI with ChatGPT). A deployer is then the user of this AI system in their own activities, for example, a marketing agency that uses ChatGPT to create content for its clients.