AI labeling requirement August 2026 for advertising and content

As of August 2, 2026, companies must disclose when content has been significantly generated by artificial intelligence, according to Article 50 of the AI Act. The ki kennzeichnungspflicht august 2026 (AI labeling requirement) affects not only large corporations but every provider and operator of AI systems acting in a professional capacity. According to the EU Commission's guidelines, there is no de minimis limit or size exemption. Anyone using AI-generated images in marketing, operating a customer service chatbot, or publishing AI-generated texts on matters of public interest is subject to these transparency rules. The regulations aim to secure consumer trust in digital content and prevent deception in commercial transactions.

Which four areas of obligation the law defines

The law divides the transparency obligation into four clear use cases covering the use of artificial intelligence in external communications. First, users of an AI chatbot or agent must be able to clearly recognize at the beginning of the interaction that they are communicating with a machine. A hidden notice in the terms and conditions or a vague designation as a "digital assistant" is not sufficient according to the Commission's guidelines. The disclosure must be proactive and unambiguous.

Second, Paragraph 2 requires machine-readable labeling for synthetic content. The open standard C2PA (Coalition for Content Provenance and Authenticity) has established itself here. This standard relies on a two-layer approach: cryptographically secured metadata is written directly into the file and supplemented by digital watermarks. This ensures the origin of an image or video remains traceable even when shared across different platforms.

Third, the law regulates the use of emotion recognition and biometric categorization. In mid-sized companies, this primarily affects HR tools in recruiting or customer service systems that perform automated sentiment analysis of callers. Anyone using such systems must inform the affected persons in advance.

Fourth, the disclosure obligation applies to deepfakes and texts. AI-generated or manipulated media that deceptively resemble real persons, objects, places, or events must be declared as artificially generated. An often-overlooked point: the offense explicitly includes objects and places, meaning AI-generated product images in online shops also require labeling. For texts, the obligation primarily applies to matters of public interest.

Why missing labels become immediately visible

A violation of the labeling requirement will be recognizable from the outside with the naked eye or through simple metadata checks starting in August 2026. Any viewer, competitor, or customer can use a missing mark as grounds for a complaint to the competent market surveillance authority, as emphasized by Online-Marketingrecht. Since the content is publicly accessible, the risk of discovery for violations is extremely high. The transparency obligations apply unchanged from the deadline, as they were not postponed by the Digital Omnibus on AI.

A transition period until December 2, 2026, exists exclusively for machine-readable labeling under Paragraph 2, and only for systems that were already on the market before August 2, 2026. There is no legally regulated transition bonus for the disclosure obligations regarding chatbots, deepfakes, and texts. According to Art. 50 Para. 5 of the AI Act, the information must be made available to users in a clear and unambiguous manner no later than the time of the first interaction.

When the editorial exception applies

The labeling requirement for texts only ceases to apply if a human reviews the content and a person or organization assumes editorial responsibility for it. A mere spelling or formal check by an employee is not sufficient to bypass the transparency obligation. The human must validate the factual accuracy and meaningfulness of the text and be able to document this in case of doubt.

Further exceptions exist for obviously artistic, creative, satirical, or fictional works. Here, the obligation is limited to disclosing the presence of AI content in a way that does not impair the enjoyment of the work. Furthermore, not all marketing texts are automatically covered. Pure advertising texts without reference to public affairs may not fall under the strict text labeling requirement, but they are still subject to general competition law transparency principles.

How systems ensure this transparency in daily operations

Legally compliant implementation requires that approval stages and transparency obligations are anchored directly in the content and communication engine. If a system scales GEO and SEO content from radar to generation across multiple AI providers in batches, governance must be built in from the start. The EU AI Act, with its risk classes, transparency requirements, and human oversight, cannot be retrofitted over an unstructured process.

In practice, this means tiered autonomy from Level 0 to Level 4. The system generates articles, product texts, or answers for customer communication, but approvals are strictly based on risk. When email, phone, messenger, and chatbot converge in a central ticket system, the engine automatically ensures that the chatbot identifies itself as AI at the first interaction. During PIM population per country and language or the updating of location pages, approval loops take effect. A human checks the suggestion, assumes editorial responsibility, and thus exempts the text from the labeling requirement. Every action, every approval, and every source reference remains traceable via an audit trail. Sensitive data stays within the defined framework, as the models are accessed via EU endpoints. In such a setup, governance is not an afterthought, but the foundation upon which automated content creation scales legally.