A growing number of websites now have a page that explains how they use AI. Nothing in the law strictly requires one. It’s still one of the easiest things you can do. It answers questions before people ask them, it gives your AI labels somewhere to point, and it shows you’ve actually thought about the subject.
Below: what to put on the page, a template to copy, and the mistakes that turn these pages into wallpaper.
A tool, not legal advice. A “How we use AI” page is a good habit. It doesn’t replace the labels and notices the EU AI Act asks for, and it isn’t a legal opinion.
Why have one at all
Article 50 of the AI Act is about telling people when AI is involved. Chatbots have to say they’re AI. Deepfakes, and certain AI-written texts, have to be labeled. The Commission’s guidelines make the point that a line buried in your terms and conditions isn’t enough by itself. The notice belongs where the content is.
So the page doesn’t replace your labels. It backs them up. A label can link to it, so a curious visitor gets the full story in one click. It can explain what a label has no room for, like which tools you use and who checks the output. And when a journalist, a customer or a business partner asks how you use AI, you can just send the link.
Think of it as an “About us” page for your use of AI.
What to put on it
Keep it short. The good ones fit on a single screen.
Open with one plain sentence that answers the question. For example: “We use AI tools to help with some of our images, texts and customer support. Here’s how.”
Then say where you use AI, in ordinary words. A small online shop might write something like this:
- Customer chat. Our chat assistant is an AI. It says so at the start of every conversation.
- Images. Some images on this site are made or edited with AI, and they have a caption.
- Articles. We use AI to help draft some articles. An editor reviews each one before it goes live.
- Product descriptions. First drafts come from AI, and our team checks them.
List only what’s true. If you don’t use AI for something, leave it out, or say plainly that you don’t.
Name the main tools. “ChatGPT”, “Midjourney” or “the Tidio AI assistant” is enough, and there’s no need for version numbers. People like knowing what’s behind the content.
Explain what visitors will actually see. For instance: “AI images have a caption that says AI-generated image. AI-assisted articles have a short note at the top.” If you use the EU icons, mention them here.
Say who checks the output, by role if not by name. “Our editor reviews every AI-assisted article before it’s published.” If you rely on the editorial-review exception for public-interest texts, the Commission’s guidelines say the reviewer needs relevant competence, and you should be able to name who holds editorial responsibility. A spell-check or a quick glance doesn’t count.
Add what you don’t do. This part does more for trust than the rest of the page put together. “We don’t use AI to write fake reviews or create photos of customers who don’t exist” tells readers a lot.
Finish with an email address for questions, or for reporting a missing label, and a “last updated” date. AI use changes fast, and a page from two years ago won’t convince anyone.
A template to copy
How we use AI
We use AI tools to help with some of our work. Here’s where, and how we keep it honest.
Customer chat. The chat assistant on this site is an AI. It tells you so at the start of every conversation and acts on behalf of [Company name]. If you’d rather talk to a person, write to [email].
Images. Some images are made or edited with AI. Those images have a caption that says so.
Articles. We sometimes use AI to help draft articles. A member of our team reviews and edits every article before it’s published and takes responsibility for it.
Tools we use. [Tool 1], [Tool 2], [Tool 3].
What we don’t do. We don’t use AI to create fake reviews, fake testimonials or photos of people presented as real customers.
Questions? Write to [email]. If you spot AI content without a label, please tell us.
Last updated: [date].
Make it yours. Change the wording, and delete any line that isn’t true for your site.
Mistakes that make the page useless
The most common one is legal boilerplate. “We may use artificial intelligence technologies in connection with certain services” tells nobody anything. Write the way you’d explain it to a customer at the counter.
Next is claiming compliance. “Fully compliant with the EU AI Act” invites exactly the question you can’t answer. Describe what you do, and let readers draw their own conclusions.
Some sites use the page instead of labels. It doesn’t work that way. The chatbot notice and the deepfake captions still have to sit on the content itself.
Others hide it, with no link anywhere. Put it in the footer, and link to it from your labels if you can.
Then there’s the stale page. You add a new AI tool in March and the page still describes last year. Put a reminder in your calendar.
And don’t promise a human you don’t have. “You can always talk to a real person” is fine only if someone is really there to answer.
Where to link it
The footer is the obvious place, next to Privacy and Terms. Link it from your AI labels too, so that clicking “AI-generated image” leads somewhere. If your chat widget lets you add a link, add one there. An “About us” page is another good spot.
How AuthorLabel helps
You get a draft in one click. In Labels & notices, click Create draft page. AuthorLabel writes a draft “How we use AI” page that lists the AI tools it found on your site, such as chat widgets and AI plugins, along with your contact details. It’s an ordinary WordPress page, so edit whatever you like and publish it when you’re happy.
It links the page for you. Tick Add a link to it in the footer and Link labels to the “How we use AI” page, and every AI caption and notice on your site will point to it.
The page stays yours. If you ever remove the plugin, the page remains.
Not sure yet which duties apply to your site? Start with the free online check.
In short
A “How we use AI” page isn’t strictly required, but it’s quick to write and it builds trust. Say where you use AI, which tools, how you label it, who checks it and what you’d never do. Keep it short, plain and dated, and link it from your footer and your labels. It backs up your labels and chatbot notice. It doesn’t replace them.
Sources
- Regulation (EU) 2024/1689 (AI Act), Article 50(1), (4) and (5).
- European Commission, Guidelines on the transparency obligations under Article 50 of the AI Act (July 2026). digital-strategy.ec.europa.eu
- Paul, Weiss, “EU finalises transparency rules for AI-generated content” (August 2026). paulweiss.com
The Commission’s guidelines aren’t legally binding.
Last reviewed: November 2026. AuthorLabel is a tool, not legal advice.