The EU mandates identification when interacting with AI from August 2

Companies using chatbots, content generators, or behaviour analysis with AI must inform users from August 2, according to the European AI Act.

Companies that use artificial intelligence in their daily operations have a new obligation since August 2: to inform users when they interact with a machine. This is mandated by Article 50 of the European Artificial Intelligence Regulation, known as the AI Act, the guidelines of which have been detailed by the European Commission.

The measure does not only affect high-risk systems, which attract much of the media attention. It also encompasses everyday uses such as virtual assistants, customer service chatbots, text, image or video generators, and tools that analyse human behaviour.

The aim, according to the European Commission, is to reduce risks such as misinformation, fraud, manipulation, or identity theft. Citizens must be able to know at all times whether there is artificial intelligence behind a conversation or content.

What companies need to do

Organizations deploying these technologies must review their systems and, if necessary, incorporate clear notices. The regulation particularly affects customer service, corporate communication, marketing, and content generation departments, where the use of AI has become widespread in recent years.

The European Commission has published guidelines to facilitate the implementation of these obligations. Companies must ensure that users are informed in a transparent and comprehensible manner, without ambiguities or fine print.

For many SMEs, the first step will be to audit which AI tools they use and in which processes they are involved. After that, they need to decide how to communicate it: a notice on the website, an automatic message in the chat, or a mention in generated content may suffice, as long as they are visible and clear.

The regulation does not require registration or prior authorisation, but it does necessitate a change in mindset: transparency is no longer voluntary but a legal requirement. Companies already using AI should review their systems before authorities begin to monitor compliance.

For users, the change means that it will soon be common to encounter notices such as "you are speaking with an automated assistant" or "this text has been generated with AI." This, according to the European Commission, will help make more informed decisions and protect against potential deception.

The next key date will be the enforcement of obligations for high-risk systems, which will come later. In the meantime, companies have time to adapt their processes and avoid surprises.