What does the AI Act mean for you if you work with websites?
The transparency requirements in Article 50 of the EU AI Act came into effect on 2 August 2026. Most organisations will hardly notice them, but if you work with website content, chatbots or digital customer interactions, it’s time to understand what they mean for you. In this article, we look at what the rules mean for those working with websites and digital communication, and what we at Epinova recommend you do about them.
What is the AI Act, in brief?
The regulation classifies AI systems according to their level of risk, ranging from prohibited applications to low-risk tools such as spam filters. Most AI tools used in web content management or marketing fall somewhere in the middle: limited risk.
This is where the transparency requirements in Article 50 come in, and they are particularly relevant if you manage a website.
What does it mean for you as an editor?
- AI-generated text must be labelled as AI-generated unless it has been reviewed by a human. Make sure you have an editorial process in place that includes human review before publication, and you can avoid the labelling requirement. For example, you could build an approval workflow into your CMS to make sure this happens.
- AI-generated images, video and audio must be machine-readable as “artificially generated or manipulated”, including hero images on your homepage.
- Manipulated content depicting real people or events (deepfake-like content) is subject to additional requirements to clearly disclose the manipulation to visitors.
- Chatbots must make it clear that the visitor is interacting with AI, unless this is already obvious.
Who is actually affected?
Most organisations are what the regulation refers to as “deployers”. You use AI systems rather than build them.
This means the requirements are less extensive than those placed on companies developing AI models. In practice, for most organisations, it is primarily about being transparent with your visitors.
Why it’s worth prioritising now
Sweden has proposed the Swedish Post and Telecom Agency (PTS) as the supervisory authority. Fines for violations of Article 50 can be as high as €15 million or 3% of global annual turnover, with lower limits for smaller companies.
In other words, it’s time to put the right processes in place and make sure they are adopted across the organisation.
How to get started
Map where AI is actually being used today; chatbots, AI-generated content in your CMS, personalisation, search. Often, there is more than you think, particularly through third-party tools.
Set up simple editorial routines for what should always be reviewed by a human, and make sure the approval is visible in the publishing workflow.
Build transparency into the user interface, for example, a clear label in your chatbot and appropriate labelling of AI-generated images.
Review your AI vendors - including chatbot and image-generation providers. Are their systems compliant?
Train editors and marketing teams in a short and practical way, using examples of what needs to be labelled.
Prioritise based on risk. Customer-facing AI features should be addressed first, while internal research tools can wait.
We can help you all the way
We see this as a technical and editorial change initiative rather than a legal project. We can help you map your current situation, build review and labelling processes into your CMS, and establish routines that editors can actually follow.
Want to know where your website stands today? Get in touch and we’ll book a no-obligation meeting to talk through your current setup.
This article is intended as an overview and does not constitute legal advice. Please consult your legal counsel or data protection officer regarding specific interpretations and requirements.
We would like to hear what you think about the blog post