Artificial intelligence has long since become part of everyday work for communications teams and agencies. Text and images are created with the help of AI chatbots, and videos are edited using AI-powered tools. Much of this happens behind the scenes. Until now, the public often had no way of knowing whether, or at what stage, AI had been involved in creating content. The EU AI Act changes this, at least in part. Since August 2, 2026, new transparency requirements have applied to AI-generated and AI-manipulated content.
For communications leaders, this raises more than the question of when disclosure is legally required. Equally important is what this new transparency means for how brands and companies are perceived. Once the use of AI becomes visible, the way content was created can itself become part of the communication.
When do companies need to disclose AI-generated content?
Not every use of artificial intelligence must be disclosed under the EU AI Act. The new requirements are particularly relevant to deepfakes, certain AI-generated texts that inform the public about matters of public interest, and direct interaction with AI systems, such as customer service chatbots.
Deepfakes are AI-generated or AI-manipulated image, audio, or video content that resembles real people, objects, places, or events and could mislead people about its authenticity. Since August 2, 2026, this content must be disclosed as such. Content depicting something that does not actually exist but is presented in a realistic way is also subject to disclosure requirements.
In communications practice, it is therefore essential to distinguish clearly between different types of AI use. Simply correcting the color or adjusting the lighting in an existing photo, for example, does not generally trigger a disclosure requirement. If, however, AI creates a realistic-looking person or situation, or alters a product to show features it does not actually have, visible disclosure may be necessary.
Do AI-generated texts and press releases need to be disclosed?
Specific requirements apply to AI-generated texts that inform the public about matters of public interest. Here, too, there is an important exception for communications teams: if content is reviewed and edited by a human and a person or organization assumes editorial responsibility for it, disclosure may not be required.
A press release therefore does not automatically have to be identified as AI-generated simply because an AI chatbot assisted with an initial draft. What matters is that people review the content editorially before publication and assume responsibility for it.
What does AI disclosure mean for brands and their communications?
The new requirements bring another question to the forefront: How will audiences respond when a company identifies content as AI-generated?
Until now, how an advertising image, social media visual, or corporate video was created has generally played little role in how audiences perceived it. A visible AI disclosure could change that. Audiences will learn not only what a company is communicating, but also something about how that communication was produced.
The impact may vary significantly from one brand to another. For a technology company, the visible use of generative AI may underscore its capacity for innovation. For a brand associated with authenticity, personal advice, or a human touch, the same disclosure may create different associations.
There is another consideration: “AI-generated” initially says little about the role the technology actually played. Did AI merely assist in creating individual elements? Was a background generated? Or did the entire piece of content come from an AI system? Disclosure therefore creates transparency about the use of AI, but does not automatically provide clarity about the extent of that use.
Communications leaders should therefore consider, from the content planning stage onward, how AI-generated content may be perceived by the relevant audience and whether that perception aligns with the message and the brand.
How should communications teams approach the EU AI Act?
When different departments within a company and external agencies work in parallel with different tools, processes can quickly become difficult to manage. Who decides whether content requires disclosure? Who reviews AI-generated texts? And who is responsible when several service providers are involved in a production?
Companies should identify their relevant AI applications and establish clear approval and disclosure processes. This includes internal guidelines as well as employee training. For text content, human review and editorial responsibility should be clearly documented. When working with agencies and other service providers, companies should also establish who is using AI and who is responsible for any required disclosure. Clear rules like these help prevent different standards from emerging within the same company.
Will “AI-generated” become part of the brand message?
How audiences will perceive content with visible AI disclosures remains to be seen. For some audiences, an “AI-generated” disclosure may eventually become relatively insignificant. Or human-created content may, for that very reason, increasingly become a mark of quality.