Β© ROOT-NATION.com - Use of content is permitted with a backlink.
The European Union has announced a new set of rules on AI transparency that companies developing or using AI systems must follow. The first rule concerns chatbots. Under Article 50(1) of the EU AI Act, users have the right to know whether they are communicating with a human or with an AI-based program, specifically with modern conversational agents or automated hotline services.
Read also: AERONAUT β everything that flies above the ground: aviation, UAVs and drones, rockets, and space
The interface must include a banner with a message such as βYou are interacting with AI,β a voice prompt at the start of the conversation, or a special AI label. The only exceptions are those chatbots where it is obvious to a βsufficiently knowledgeable, observant, and discerningβ person that they are interacting with artificial intelligence β for example, an NPC in a video game.
Read also:Β How to Make an AI Explainer Video: A 7-Step Workflow
The EU also requires that AI-generated content be properly labeled using means such as machine-readable metadata, cryptographic origin markers, or digital watermarks (e.g., SynthID), embedded directly into the systemβs output. An exception is provided only for cases where the system makes minor changes to the source material β for example, as done by systems that correct grammar or upscaling images without altering their original content.
If an AI system can identify users by analyzing biometric features β such as facial expressions or voice tone to determine emotions β or classify people into specific demographic categories, it must disclose this before the system is used, for example, via a pop-up window or a clearly visible physical notice. However, there are exceptions here as well β in particular, law enforcement agencies may not disclose the use of AI for surveillance during active criminal investigations.
Read also:Β Everything You Need to Know About Moonshot AI’s Kimi K3: How Chinese Startup Challenged AI Giants
Finally, EU representatives note that if content created or modified using AI closely resembles a real person, object, place, or event, it must include a label informing viewers of its artificial origin. In creative contexts, such as when AI is used to make an actor in a movie look younger or to create clearly satirical music, the labeling must be implemented in a way that does not detract from the viewing experience or the intended interpretation of the work.
Companies that fail to comply with these requirements may face a fine of up to β¬15 million or 3% of their global annual turnover, whichever is greater.
Read also:Β The Age of Realistic Science Fiction: Google’s New AI Gives Androids Perfect Balance and Human-Like Agility