The EU AI Act has been in effect since August 2024 and is being implemented in phases. The next phase is a major one: Starting August 2, 2026, the Transparency obligations from Article 50 for providers and operators of AI systems—including companies that merely use tools such as ChatGPT or Claude. At the same time, the EU has introduced the Digital Omnibus In June 2026, several other deadlines were pushed back. The deadline for labeling remains unaffected, which is exactly why it’s worth taking a closer look at the details now.
What Article 50 requires starting August 2, 2026
Article 50 governs situations where people encounter AI without necessarily noticing it. Anyone who Chatbot Any company that uses AI in customer service or on its website must, going forward, inform users that they are communicating with a machine. This information must be provided no later than the first interaction, in a clearly visible manner and in understandable language. The only exception applies if the use of AI is obvious to an average informed user.
The second major construction site is the Labeling Requirement for synthetic content. Providers of systems that generate audio, images, videos, or text must label the results as AI-generated in a machine-readable format, for example by Watermarking such as Google's SynthID or via embedded C2PA metadata, as used by OpenAI in generated images. Operators of Deepfakes must additionally disclose visibly that the content was artificially generated or manipulated. You can find the complete text of the regulation, including all exceptions, at EUR-Lex.
Who is affected by the obligations and what penalties apply for violations
The AI Act distinguishes between two roles: providers develop and distribute AI systems, while operators use them in a professional capacity. The transparency requirements apply to both, which is why small and medium-sized businesses must also address this issue, as the IHK emphasized. Who Generative AI whether used for product images, short video clips, or entire campaigns, is legally considered an operator and bears its own responsibilities. For AI texts that inform the public about matters of public interest, the regulation requires disclosure unless a human demonstrably bears editorial responsibility.
Brussels is hitting hard with sanctions: violations of Article 50 can be punished with up to 15 million euros or 3 percent of the worldwide annual turnover, whichever amount is higher. That's a significant figure. For your daily work with AI in marketing This means documenting processes, defining responsibilities, and considering labeling from the outset. Those who set this up cleanly turn a duty into a signal of trust towards customers and business partners.
The Digital Omnibus postpones other deadlines
The EU’s Omnibus Package, which was adopted in June 2026 and realigns the timing of several components of the AI Act, has recently caused further confusion. The European Parliament approved it on June 16, the Council followed suit on June 29, and the details of the legal framework will be published by the European Commission. In particular, the obligations for High-Risk Systems, as Article 50 is set to enter into force. The reasons for the postponement are the lack of technical standards and concerns about the competitiveness of European companies, which result in the following timeline for you:
- August 2, 2026: The transparency and labeling requirements under Article 50 are binding on providers and operators.
- December 2, 2027: The obligations for standalone high-risk systems under Annex III will apply, with August 2, 2026, originally planned for this.
- August 2, 2028: AI in regulated products listed in Annex I is the final phase; the previous deadline was August 2027.
This is how you get your company fit in time
Sounds like a lot of work? With a structured roadmap, you’ll be able to get a handle on the requirements quickly, and much of it is part of professional AI implementation anyway. The more autonomous the systems become, the more important the foundation becomes: A AI Agent are now taking over entire workflows, and models such as Claude Fable 5 They handle tasks that used to require entire teams. If you’d like some help with this, you might know of a digital agency in Düsseldorf that’s quite knowledgeable about AI—these five steps will get you ready by the deadline:
- Conduct an AI inventoryCaptures which tools are running in the company and whether you are acting as the provider or operator for each. This role distribution determines which obligations specifically apply to you.
- Upgrading Chatbots: Include a clear indication that users are communicating with an AI, preferably right at the start of the conversation. A short sentence in the first dialog window is sufficient to meet this requirement.
- Automate LabelingCheck if your generators are embedding machine-readable watermarks and add visible cues for deepfakes. For tools without built-in watermarking, you should switch to alternatives.
- Train the teamThe requirement for AI competence is expressly enshrined in the AI Act; regular training directly contributes to your Compliance one.
- Build documentationKeep track of who created which content with which tool; this greatly simplifies any later review.
FAQs on the EU AI Act
Article 50 of the EU AI Act regulates the setting up of an AI Office.
Article 50 contains the transparency obligations of the EU AI Act. It obliges providers and operators to disclose the use of AI, for example in chatbots, AI-generated content, and deepfakes. The rules will be binding throughout the EU from August 2, 2026. The goal is for people to be able to recognize at all times when they are interacting with AI or seeing AI-generated content.
Who will be affected by the transparency obligations starting in August 2026?
This applies to providers of AI systems as well as operators—that is, companies that use AI for business purposes. This means that small and medium-sized enterprises are also subject to these obligations as soon as they operate chatbots or create and publish content using AI. Article 50 applies regardless of revenue or company size; the only exception is purely private use.
How must companies label synthetic content?
AI-generated audio, image, video, and text content must be labeled in a machine-readable format, such as through watermarks or metadata. In the case of deepfakes, visible disclosure is also required so that users can immediately recognize the artificial origin. The provider of the respective AI system is responsible for the technical labeling.
What penalties apply for violations of the EU AI Act?
Violations of transparency obligations are punishable by fines of up to 15 million euros or 3 percent of global annual revenue, whichever is higher. Article 99 of the regulation is decisive. For small and medium-sized enterprises, the lower of the two amounts applies as the upper limit.
Does the labeling requirement also apply to marketing texts?
Visible disclosure is mandatory for texts, especially when they inform the public about matters of public interest and the editorial control lies solely with the AI. For advertising texts, which are reviewed and approved by a human, machine-readable marking by the provider of the AI tool is sufficient. You can still use voluntary labeling; many customers appreciate this transparency.
