Do you use artificial intelligence to write texts, create images, or edit videos for your business or association? If so, there are new rules from the EU that you need to know.
On August 2, 2026, the next wave of the EU’s AI Act comes into effect. It might sound like dry legalese that only affects tech giants, but it has a direct impact on your business or association.
Artificial intelligence (AI) is a fantastic gift for those of us working with tight budgets and limited resources. It enables the local craftsman, the sports association, and the small webshop to create professional content in record time.
However, as AI content becomes harder to distinguish from reality, the EU has introduced new guidelines in their sweeping “AI Act.” The purpose is to protect trust in what we read and see online.
As a sender of marketing, it is now your responsibility to inform your audience if your content is created or manipulated by a machine. In other words, you must strengthen transparency in a time when the authenticity of digital content cannot be trusted.
When You MUST Disclose Your Use of AI
You don’t have to slap an AI label on everything you touch just because you used AI for spellchecking, for example. The rules are specifically aimed at two main categories:
There are specifically two areas in your daily marketing where you need to be transparent about your use of AI.
1. “Deep Fakes” (Images, Video, and Audio)
If you use AI to generate or manipulate images, videos, or audio that resemble real people, places, objects, or events so that it appears authentic, it must be labeled.
If you ask an AI program to create an image or a video that is indistinguishable from reality—for example, pictures of smiling “customers,” or fake scenarios from a store or of your products—it must be labeled. Users must not be led to believe a fake image is real.
2. Published Texts and General Information
The EU uses the term texts on “matters of public interest.” Here you might think: “Well, I don’t write about high politics or global crises. I just make Facebook posts about tax deductions for home improvements, guides for the perfect lawn, or news about the local football team.”
But in the eyes of the EU, “public interest” can easily cover the perfectly ordinary, informative texts you post on your website or social media. It applies broadly when you publish information intended to inform the public (including your customers and your local community).
If you ask ChatGPT to write an informative article, a guide, or a professional post, and you hit “Publish” without having read through the text yourself, edited it, and taken responsibility for it, then the text must be labeled as artificially generated.
This also applies if you have a chatbot on your website or social media.
Many small businesses today use an automated chatbot to answer simple customer queries outside of opening hours. It is an excellent service, but the rules require that the user is fully aware they are speaking to a machine.
Make sure the chatbot always introduces itself correctly. The first message could, for example, read: “Hello! I am your digital assistant (an AI robot). How can I help you today?”
What Are You Exempt From Labeling?
Fortunately, the rules are designed so you don’t have to put warning triangles on everything you do. You can, for example, completely avoid labeling requirements for text material simply by using your common sense.
The rules state that if the text has undergone human editing and oversight, and you take editorial responsibility for the publication, the requirement is waived. So, feel free to use AI as your personal copywriter to get good ideas and drafts on the table—but always read through the text and leave your own human fingerprint on it before sharing it with the world.
Therefore, as a rule of thumb, you do not need to label your content if:
- You use AI to correct typos or improve the language in a newsletter you have written yourself.
- You use AI to brainstorm ideas for your next campaign.
- You ask AI to write a draft for a product or SEO text, which you then edit and publish yourself.
How to Label Your AI Content Correctly
When labeling your content, it must be done in a way that is clear and easily distinguishable to the recipient the very first time they see it.
The EU has developed specific design requirements, and to make it easy, they have designed official icons you can use.
1. Use the Right Words and Icons
It is a requirement that your labeling clearly includes the abbreviation “AI.” Additionally, it is recommended as a best practice to combine “AI” with more specific text, as user tests show this provides greater clarity for the recipient.
- Use “AI GENERATED” for content that is 100% created by artificial intelligence.
- Use “AI MODIFIED” for content where you have used AI to alter or manipulate a portion of an existing image or video. However, this only applies to minor edits, such as color adjustments, etc. You must not add new elements to images; otherwise, it must be labeled “AI GENERATED.”
2. Placement of the AI Label
- Images:
The icon must be placed in a visible location, for example in the top right corner of the image. The label must be permanent—and not just visible on mouseover. Likewise, you cannot make it faded and only highlight it on mouseover. - Video:
At the beginning of the video, it must be displayed that the video contains AI-generated or AI-modified content. Furthermore, it must be shown several times throughout the video—and if there are interruptions (e.g., ads), it must be shown again after each interruption. However, the EU recommends displaying the icon during all segments of the video that include AI-generated or AI-modified material. - Placement in Text:
If your text falls under the labeling requirement (i.e., without human editing), you must place the icon or label at the top of the text, right below the headline or at the beginning of the article. - Audio and Podcasts:
If an icon cannot be seen, for example in a podcast, you must include a short and easily understandable spoken message at the beginning of the broadcast stating that the audio is AI-generated or manipulated. If it is a long audio file, you should repeat the warning throughout.
Download the Official AI Icons
To make it as easy for you as humanly possible, the EU icons are available for you right here on this page. They are ready to download and place over your images and videos in your preferred editing program (e.g., Canva).
The EU’s design is made to be used by anyone, and they are available with black, white, and transparent backgrounds, ensuring they are always visible on your content.
👉 YOU CAN DOWNLOAD THE AI ICONS HERE
Note that if you use the icon that simply says “AI,” it must be accompanied by a text element indicating whether it is generated or modified by artificial intelligence.
Beware of the Trap of AI Images of Your Products
Right now, I am seeing a clear—and slightly concerning—trend when scrolling through my social media feed. More and more local restaurants, cafes, and takeout places have started using AI to generate pictures of their food. This results in completely flawless, extravagant burgers and magical pizzas that either look like a million bucks or like plastic food from Fisher-Price.
I fully understand that it is incredibly tempting. It saves time, and it looks fantastic on screen. But it is a dangerous path to take for your business.
Here is why you should be careful letting AI “cook the food” in your marketing—and what you should do instead:
- Broken Trust Costs Customers:
When a guest visits your restaurant based on a hyper-realistic AI image, they expect to be served exactly what they saw on Facebook. If the reality on the plate doesn’t match (and it rarely does with AI food), you create an unnecessary disappointment. Disappointed customers don’t return, and at worst, they leave bad reviews. - Beware of Marketing Laws:
Even if you comply with the EU’s new AI rules and nicely place an “AI GENERATED” label on your image, you are moving in a dangerous gray area regarding marketing laws. If you show a product that differs significantly from the actual product the customer ends up buying, it can be considered misleading marketing. - Authenticity Beats Perfection:
As a small, local business, your greatest and most important strength is your authenticity. Customers in your local area want to see your food, made by your hands in your kitchen. A completely normal, honest picture of a delicious dish you just plated builds far more credibility than a soulless and plastic-like AI image.
How to Use AI for Your Creatives
Does this mean you should drop AI entirely? Not at all! You just need to use it strategically where it makes sense and creates value for you:
- For Cleaning Up Images:
Use AI tools on your smartphone (such as “Magic Eraser” or Canva’s AI features) to remove a messy cable, a dirty cloth, or a trash can in the background of your real food picture. - For Catchy Text:
Shoot an honest picture of a product with your phone, and let ChatGPT help you write a fun and engaging text for your post, so you don’t have to stare blindly at a blank text document. - For Mood and Graphics:
Use AI to create beautiful frames, backgrounds, or graphics for your menu or promotional signs, where the product itself isn’t the focus, but where it just needs to look inviting.
Maintain Your Common Sense
The most important thing you can take away from these new rules is to build trust with your followers and customers. Transparency about your use of AI is not a weakness—it shows that you are a modern, honest business keeping up with the times. As long as you include your own critical thinking and your own human touch in the process, you are already well on your way!
FAQ about AI in Marketing
No, the EU’s AI Act generally does not require you to label standard text posts if the AI merely assisted you in drafting them. The transparency requirement applies primarily to interactive chatbots and photorealistic image/video material. However, it is always good practice to be honest about your working methods.
Failure to comply with transparency requirements can, in extreme cases, result in injunctions and fines from the authorities. For small businesses, however, the greatest risk is often the loss of customer trust if they discover they have been misled by fake images.
Yes, absolutely! You are welcome to use AI tools to create graphics, logos, and images for your website or social media. You just need to ensure it is clearly stated in the post or on the image that it was created with artificial intelligence if the image resembles a real photograph (so-called deepfakes).
