AI Act and websites: what freelancers and SMEs using AI need to know

Published by Incomedia in Management & Law · Monday 21 Sep 2026 · 13 minutes

Artificial Intelligence has now become part of the day-to-day management of websites too.

It can help you write the content of a page, prepare articles, generate images, describe the products in your e-commerce store, or answer visitors through a chatbot.

With the application of the European AI Act, questions naturally arise, such as: “Do I need to do anything too? Do I need to change something or declare every time I use AI?”

The answer depends on how you use AI, but in many cases, there is no need to worry. For most individuals, small businesses, freelancers, and associations, the obligations are much simpler than you might think.

Let’s make one thing clear right away: you can continue to use AI tools, but it is important to check what you publish and recognize the situations in which you need to be more transparent.

So, let’s take a look at what is really changing and what you need to do in the different situations that may arise.

What is the AI Act and why is it being discussed now?

The AI Act is the European regulation that governs the development and use of artificial intelligence systems. It entered into force on August 1, 2024, and from August 2nd, 2026, the transparency rules also apply to, among others, chatbots, deepfakes, and certain content generated or manipulated by artificial intelligence.

The AI Act does not treat all uses of AI in the same way. Instead, it follows a risk-based approach, distinguishing between prohibited uses, high-risk systems, systems subject to specific transparency obligations, and uses that pose minimal or no risk.

The rules therefore become stricter when a system can have significant consequences for people's safety, rights, or lives, for example in recruitment, access to credit, healthcare, or education.

Many of the tools commonly used to generate ideas, text, images, or web pages do not fall into the high-risk categories. This does not mean, however, that AI can be used without any precautions.

For anyone who creates or manages a website, the key practical points to keep in mind are:

Transparency: in certain situations, people must be able to recognize an interaction or content generated artificially.
Human oversight: before publishing AI-generated content, it is good practice to review it.
Responsibility: responsibility for what is published on the website remains with the person managing it.
Informed use: anyone using AI should be aware of its main limitations, such as the possibility of generating inaccurate or fabricated information.

The practical rule is simple: AI can help you create. You check and decide.

Does the AI Act also apply to people who use AI to create a website?

The AI Act does not only concern large technology companies that develop artificial intelligence systems.

Among other things, the regulation distinguishes between those who develop and place an AI system on the market and those who use it as part of their professional activities.

The latter are referred to by the regulation as deployers, which can be more simply understood as professional users.

A freelancer who uses an AI assistant to write the content for their website, a shop that uses artificial intelligence to prepare product descriptions, or a small business that integrates an AI system may therefore fall under the category of professional users.

This does not mean that they automatically have to deal with complicated procedures. The obligations depend on the system being used, its function, and how it is deployed.

So, the question to ask is not simply:

Have I used artificial intelligence?”

But rather:

How have I used it, and what impact could it have on the people visiting my website?”

What about personal websites?

If you use AI to create a personal website, prepare a text, or come up with ideas for a non-commercial project, you are generally not the primary target of the business obligations set out by the AI Act.

Even for a personal website, however, some common-sense rules still apply:

always check the information generated;
do not publish false or offensive content;
do not use images or text that infringe other people’s rights;
do not enter passwords, confidential information, or unnecessary personal data into AI tools.

AI can be a great assistant, but it is not an infallible source.

Where can AI be present on websites?

When you use artificial intelligence to create a website with WebSite X5 or to manage it over time, it can be useful to distinguish between two situations.

AI used behind the scenes

In many cases, artificial intelligence is used while creating or managing a website, without directly interacting with visitors.

For example, you may have used an assistant to:

suggest the structure of the website;
come up with a page title;
prepare a first draft of a text;
rephrase or translate content;
create an illustrative image;
write a product description;
develop ideas for blog articles.

In these situations, AI is a support tool. The content passes through you before it is published.

The most important thing is to remain in control of the result: check the information, correct any errors, adapt the tone, and make sure that what you publish genuinely represents your business.

AI interacting with visitors

The situation is different when an AI system communicates directly with people.

This may happen, for example, with:

a chatbot that answers questions;
a virtual assistant for customer service;
an AI agent that provides personalized responses in real time.

In this case, the visitor is not simply receiving content that was prepared in advance: they are interacting directly with an AI system.

And this is where transparency becomes particularly important.

Diagram showing how AI is used on websites: behind-the-scenes content creation and direct interaction with visitors

At this point, the picture is clearer: using AI to create or manage a website can mean very different things.

It can be a tool that helps you work behind the scenes, or a system that interacts directly with visitors. It can be involved in writing text, creating images, or managing a conversation.

And depending on how it is used, the precautions you need to take can also change.

So, let’s look at the most common scenarios: what data you should share with AI tools, how to handle generated text and images, and what to do when a chatbot interacts directly with website visitors.

Before writing a prompt, check the data you’re sharing

When you use an AI service, anything you enter in the prompt is processed according to the provider’s terms and policies.

Before sharing personal data, documents, or confidential information, check how they will be handled and limit the information you provide to what is genuinely necessary.

In particular, be careful before entering:

passwords and login credentials;
payment card details;
personal documents;
health information;
customer lists;
employee or collaborator data;
contracts and confidential documents;
confidential business information.

If you need to work on a real-life case, remove names, addresses, and other unnecessary identifying details whenever possible.

This consideration does not depend solely on the AI Act. When personal data is involved, the GDPR and other data protection regulations continue to apply.

In other words, using an AI assistant does not automatically mean that it is appropriate to share any information you have available with it.

Diagram showing three use cases for AI on websites: text to be reviewed, images and deepfakes to be checked, and chatbots to be clearly identified as AI systems.

AI-generated text: do you always have to disclose it?

This is probably one of the most common questions.

The answer is: no, you do not automatically have to disclose every use of artificial intelligence when writing a text.

Imagine asking an AI assistant to prepare the first draft of your “About Us” page. After receiving the text, you check the information, rewrite some sections, add your company’s story, and personally approve the final result.

Artificial intelligence has contributed to the writing, but the final content has been genuinely reviewed and published under the responsibility of a person or organization:

a product description reviewed by the company;
a translation checked before publication;
a blog article reviewed for its content and sources;
a service page rewritten to make it clearer;
a title or call to action suggested by an assistant.

Article 50 of the AI Act instead provides a specific transparency rule for text generated or manipulated by artificial intelligence and published with the purpose of informing the public about matters of public interest.

This may include, for example, political, social, economic, health, or environmental issues that are relevant to public debate.

Even in these cases, however, the obligation to disclose the artificial nature of the text does not apply when the content has undergone a process of human review or editorial control and a natural or legal person assumes editorial responsibility for its publication.

It is important to stress, however, that genuine review goes beyond simply correcting grammar or changing a few words. It means, for example:

checking the accuracy of the information;
verifying the sources;
assessing whether any important elements are missing;
modifying ambiguous or misleading statements;
consciously deciding whether the text can be published.

An example

You asked AI to write an article about the new tax rules for freelancers.

If you copy and publish it without checking the data, deadlines, and sources, you risk spreading inaccurate information.

If, on the other hand, you check the relevant legislation, update the data, assess the sources, modify the text where necessary, and take editorial responsibility for the publication, your involvement goes beyond simply making formal corrections. This is consistent with the human review and editorial control exemption provided for in Article 50 of the AI Act.

The most useful rule to remember is: AI can help you write, but what you publish should always be reviewed and approved by you.

AI-generated images, videos and audio: when is greater transparency required?

The same principle applies to images: there is no general rule requiring every piece of content generated by artificial intelligence to be accompanied by a label such as “AI-generated image.”

However, particular attention should be paid to deepfakes.

Under the AI Act, a deepfake is image, audio or video content generated or manipulated by artificial intelligence that resembles existing people, objects, places, entities or events and could falsely appear authentic or genuine. When content constitutes a deepfake, a professional user must disclose that it has been artificially generated or manipulated.

Let’s look at a few examples.

A clearly artificial illustration

You create a stylized illustration of a person sitting at a computer to use as the opening image for an article.

The audience has no particular reason to interpret it as a photograph of a real person or as documentation of an event that actually took place.

The fact that the image was created with AI does not automatically mean that it should be treated as a deepfake. Under the European Commission’s guidance, the context, the degree of resemblance to reality, the intended audience, and whether the content could falsely appear authentic or truthful are relevant when determining whether the deepfake transparency obligation applies.

A photorealistic image of a non-existent person

You generate a photorealistic portrait of a doctor and add it to the “Our Team” page without making it clear that the person is not a real professional.

Visitors could reasonably believe that this person actually works for the company. This is therefore a situation that requires much greater care and transparency.

An event that never happened

You generate a realistic image of a well-known personality using your product, even though this person has never actually done so.

The image could be interpreted as authentic and falsely attribute a behavior or endorsement to that person. In such a situation, the artificial nature of the content must be made clear.

An AI-generated product image for an e-commerce store

You generate a photorealistic image of a product that does not exactly match the one you are selling.

Even when the case does not fall under the definition of a deepfake, you still need to avoid creating misleading expectations about the product’s features, appearance, or use.

For clearly artistic, creative, satirical, or fictional works, the AI Act instead allows information about the artificial nature of the content to be provided in an appropriate way without compromising its normal enjoyment.

Once again, the point is not to indiscriminately label everything created with AI, but to prevent artificial content from being mistaken for something authentic.

Chatbots and virtual assistants: visitors should know who they are talking to

If your website has an AI system that interacts directly with visitors, people must be able to understand that they are talking to an artificial intelligence, unless this is already obvious.

This information must be provided from the beginning of the first interaction, in a clear and distinguishable way.

A simple wording could be: “Hi, I’m the AI-powered virtual assistant for [company name].”

There is no need to turn the greeting into a long technical notice. The goal is simply to prevent the person from believing they are talking to a human being.

It is also important to distinguish this case from AI used while building the website.

If you use a WebSite X5 assistant to prepare a page, the visitor is not directly interacting with that system: they see the content you have decided to publish.

A chatbot that communicates with visitors in real time is a different situation.

In brief: what to do in the most common situations

Let’s summarize some of the most common situations.

You use AI to come up with ideas or titles → Personally evaluate and select the suggestions
You generate the first draft of a page → Check the information, tone, and consistency before publishing it
You write a product description with AI → Verify features, measurements, availability, and commercial claims
You translate a text with an assistant → Check that the meaning remains accurate
You generate a decorative image → Assess whether it could be mistaken for an authentic representation of a person or an event
You publish a deepfake → Clearly indicate that the content has been generated or manipulated with AI
You use a chatbot on your website → Make sure visitors know they are interacting with an AI system
You publish AI-generated text on a matter of public interest → Carry out a genuine editorial review or check whether its artificial nature needs to be disclosed
AI is used by employees or collaborators → Provide practical guidance on the tools, risks, and checks to be carried out

This table does not replace professional assessment in more complex cases, but it can provide a useful starting point for identifying situations that require greater attention.

How these principles apply to WebSite X5 assistants

WebSite X5 integrates AI tools that can intervene at different stages of website creation. Let’s see how to apply what we have discussed in practice.

MagicText

MagicText, WebSite X5’s AI assistant for text, can help you create the text content for your website: titles, paragraphs, product descriptions, or articles. It can also rework existing text, shorten it, expand it, simplify it, optimize it, or translate it.

The result remains editable, and you decide whether and how to use it.

Before publishing it, treat it as you would a colleague’s contribution:

check that the information is accurate;
adapt it to your business and your audience;
verify names, data, prices, and features;
remove any unnecessary references;
proofread it to make sure it really communicates what you want to say.

You do not automatically have to disclose that you used MagicText to prepare a standard website page. What matters is the type of content and, in the cases covered by the AI Act, how it was reviewed before publication.

MagicSite

With MagicSite you can create a website with WebSite X5’s AI starting from a conversation: it asks you questions about the type of website you want to create, your goals, and the characteristics of your business.

Based on your answers, it suggests the style and page structure and then generates a complete project with content that you can open and customize in the WebSite X5 editor. Texts, images, colors, fonts, layouts, and features can all be modified before publication.

This is also important from the perspective we are discussing: MagicSite creates a starting point. You control it, customize it, and decide what to publish.

Using MagicSite to create your project therefore does not mean that website visitors are directly interacting with artificial intelligence.

Once again: AI suggests. You check and decide.

AI users also need to understand its limitations

The AI Act also focuses on AI literacy, meaning the development of the knowledge and skills needed to use these systems responsibly and with awareness.

Those who use AI systems professionally must take measures to support the development of the skills of staff and other people who use AI systems on their behalf, taking into account the context, the tools, and the experience of the people involved.

For a small business, this does not necessarily mean organizing complex training courses.

There is no one-size-fits-all training program, nor is a specific certificate or particular organizational structure required. Initiatives should be proportionate to the specific situation.

If you, your employees, or your collaborators mainly use AI for your website and communications, you can start with a few shared rules:

do not automatically publish what AI generates;
check data, names, numbers, and sources;
remember that the system may produce inaccurate or fabricated information;
be careful before entering personal or confidential data;
pay particular attention when reviewing content that concerns real people;
know when a chatbot or AI-generated content needs to be identifiable as such;
seek expert advice when AI is used in particularly sensitive contexts.

Even a short internal document containing these guidelines can be a good starting point.

A quick checklist before publishing

Have you created a website, page, or piece of content with the help of artificial intelligence? Before publishing it, try answering these seven questions.

1. Do I know which AI tools were used?

Consider both the assistants integrated into the software and any external services used for text, images, translations, or other tasks.

2. Is the information accurate and up to date?

Pay particular attention to names, prices, product features, statistics, quotes, and technical information.

3. Have I avoided unnecessary personal or confidential data?

Check what you entered in the prompts and what appears in the final content.

4. Do the texts and images respect other people’s rights?

Pay particular attention to real people, trademarks, protected works, and content that could falsely attribute statements or endorsements.

5. Could the content mislead someone about its authenticity?

Ask yourself whether an image, video, voice, or text could be interpreted as an authentic representation of a person, product, or event.

6. If I use a chatbot, is it clear that it is an AI system?

Visitors should be able to understand this from the beginning of the interaction.

7. Have I carried out a proper final review?

Don’t just check that the content “sounds good.” Make sure it is accurate, relevant, and suitable for publication on your website.

If you have completed all the necessary checks, you are already on the right track.

Hai usato l’AI per il tuo sito? Non devi smettere

The AI Act is not an invitation to abandon AI assistants or give up the benefits they can offer.

For those who create and manage a website, the main message is much simpler: you can continue to use artificial intelligence, but you should do so responsibly and with awareness.

If you have used an assistant to come up with ideas, improve a text, create an illustrative image, or prepare a first draft, you normally do not need to add a disclosure to every page.

You need to pay more attention when AI interacts directly with visitors, when you publish artificial content that could be mistaken for authentic content, or when you use AI-generated text to inform the public about matters of public interest.

This is also the principle behind the use of the assistants integrated into WebSite X5: artificial intelligence can make your work faster and easier, but you remain in control of the result and decide what to publish.

Want to put these principles into practice on your next project? Get started with WebSite X5 for free and create your website while always staying in control of the result.

This article is for informational purposes only and does not constitute legal advice. Obligations may vary depending on the type of business, the AI system used, and how it is deployed. In complex cases, or when artificial intelligence is used to make decisions that affect people, it is advisable to consult a qualified professional.