Obveza transparentnosti od kolovoza EU uvodi oznaku AI: što se sada mijenja

Starting on 2 August, all AI-generated content used for professional purposes within the European Union will need to carry a clear label. This new transparency requirement marks the implementation of the EU’s AI labeling system, designed to enhance trust in digital content and reduce the risk of misinformation. The regulation applies to images, text, and audio created or manipulated by artificial intelligence, particularly when these materials serve commercial, journalistic, or informational functions. The rule covers deepfakes, AI-generated or altered visual and audio content that could be mistaken for genuine material, as well as fully automated texts produced by AI without human oversight, such as certain public announcements. Users must also be made aware when they interact with chatbots, ensuring they recognize they are communicating with machines rather than humans. These measures aim to empower users to better assess the authenticity and origin of information they encounter online. The labels themselves will take one of three forms: circular stamps bearing the words “AI,” “AI generated,” or “AI modified.” While there is no formal standard dictating the exact design of the label, clarity and visibility are paramount. In addition to visible labels, starting in December, AI-generated content will also require a digital watermark, which will be implemented by AI service providers such as ChatGPT and Google’s Gemini. These watermarks will help verify the content's origin and ensure compliance with the new rules. The obligation to label AI-generated content applies to anyone who professionally shares such material, including businesses, influencers, and media outlets. Private individuals uploading AI-generated content to personal social media profiles are generally exempt, though platform-specific transparency policies may still apply. Organizations based outside the EU, including those in Switzerland, are required to comply if their content is directed toward EU audiences. Enforcement of the regulations falls under the responsibility of national authorities, with the German Federal Network Agency overseeing compliance in Germany. Violations can result in fines up to three percent of annual revenue, serving as a deterrent against non-compliance. The EU has emphasized that this regulatory framework is part of broader efforts to establish ethical guidelines for AI usage and protect consumers from misleading or harmful content. As the deadline approaches, major tech companies and platforms are preparing to implement the necessary changes. Some have already introduced their own labeling systems, aligning with the EU’s proposed standards. Industry experts suggest that while the initial focus is on compliance, long-term implications include greater accountability for AI developers and more informed consumer behavior. The success of the initiative will depend on how effectively these measures are enforced and how widely they are adopted across both digital platforms and traditional media channels.