What Are the New Rules?
Under Article 50 of the EU's comprehensive AI Act, any company whose services reach users in the EU must now ensure people know when they are interacting with artificial intelligence. This applies to chatbots, which must identify themselves as non-human,
as well as to synthetic media. Any AI-generated image, audio, or video that looks authentic—often called a deepfake—must be clearly labeled. Similarly, AI-written text on matters of public interest, such as news or political commentary, must also be disclosed, unless a human editor has substantively reviewed it and assumes editorial responsibility. For creators of AI models, the rules go a step further, requiring them to embed machine-readable watermarks or metadata to help detect synthetic content. The regulations apply globally, meaning any company, whether in India or the US, must comply if its AI-driven content is viewed in the EU. Failure to do so can result in staggering fines of up to €15 million or 3% of a company's global annual turnover.
The Goal: A Stand Against Disinformation
The core purpose of these rules is to protect consumers and democracy itself. In recent years, the explosion of generative AI has made it incredibly easy to create convincing but fake images, voice clones, and videos, leading to a surge in sophisticated disinformation campaigns and fraud. European regulators believe that giving citizens the ability to distinguish between authentic and artificial content is a fundamental requirement for a healthy information society. By forcing transparency, the EU aims to make it much harder for malicious actors to spread fake news, manipulate public opinion, or impersonate individuals without the viewer’s knowledge. The move is part of the EU AI Act, a broader legislative package that is being rolled out in stages and is considered the world's first comprehensive law on artificial intelligence.
The 'Cookie Banner' Problem
While the goal of transparency is widely supported, the implementation poses a significant test. Critics, including some tech industry associations, fear the rules are too broad and will force companies to label almost all AI-assisted content, not just the deceptive kind. The concern is that this will lead to “label fatigue,” where users become so accustomed to seeing AI warnings that they begin to ignore them entirely, much like many now reflexively click through cookie consent banners. This could undermine the entire purpose of the regulation, failing to protect anyone while adding a significant compliance burden on businesses, especially smaller ones less versed in AI law. The challenge lies in finding a balance that flags genuinely manipulative content without overwhelming users with constant, and ultimately meaningless, disclosures.
Ripples Beyond Europe
Just as the EU’s General Data Protection Regulation (GDPR) became a de facto global standard for data privacy, experts anticipate these AI labeling rules will have a similar “Brussels effect.” Multinational technology companies often find it easier to apply the strictest standard across all their operations rather than creating different versions of their products for different regions. As a result, users in India and around the world will likely start seeing more AI labels on social media, in advertising, and across the web as companies adapt to the EU’s framework. Major tech firms had already begun rolling out their own transparency measures, but the AI Act formalizes these efforts under the force of law, creating a common benchmark for the entire industry.
