Disclosure of AI interaction is a transparency requirement according to which it must be clear to a person that they are communicating with a machine, or that the content they are seeing is artificially generated. In practice, it mainly concerns corporate chatbots and voice assistants, which must be clearly labelled, and synthetic content, where labelling of generated or manipulated material is expected. Cases where this is obvious from the context are an exception. Compliance is cheap, though: visible information right at the start of the conversation, the option to switch to a human, a note under the image or video. The risk does not lie only in the sanction – a customer who discovers after twenty minutes that they were chatting with a bot reacts worse than if they had known from the start. Check the scope of the obligations in the current text of the regulation.
See also: Human oversight of AI, Internal AI usage policy, AI risk classification.
