What is the EU AI transparency code? | Explained

Image used for representational purposes only. | Photo Credit: Getty Images/iStockphoto The European Union is set to make labelling of AI-generated content mandatory from August 2, in a move to ensure transparency and enable human users to distinguish manipulated content online. Falling under the purview of Article 50 of the EU Artificial Intelligence Act (AIA), which sets a framework for the use and supply of AI systems within the bloc with a risk-based categorisation, the ‘Code of Practice on Transparency of AI-Generated Content’ stipulates certain obligation for AI developers and providers to maintain the integrity of the information ecosystem, with administrative fines up to €15 million for non-compliance.What are the contents that need to be labeled under the new code? The European Commission has summarised the scope of the transparency code into two broad categories of online content: deepfakes and manipulated text. Deepfakes refer to images, audio, or video resembling existing persons, places, or events that are generated (or manipulated) with AI and would appear to a human user as authentic. Manipulated text includes that published to pass information to the public that has not undergone human review. The code requires AI system developers and content providers to display prominent, distinguishable signs that such products are artificial.What are the exceptions to the new code? While the code seeks to make digital spaces trustworthy for human users, it has given certain leeway for the use of AI, especially for creative content. For deepfakes that are part of products that are evidently artistic, satirical, or fictional, the code has simplified transparency obligations, stipulating appropriate disclosure without hampering the display of the work. The disclosure obligation is also done away with for content “authorised by law to detect, prevent, investigate, or prosecute criminal offences”. AI-generated text that has undergone human review or is under editorial control with a legally liable person holding responsibility for it, is also exempt from the obligation.How should an AI-generated content be labeled under the code? The EU has listed out three kinds of icons to be displayed with deepfakes and manipulated text for compliance with the AIA: a basic ‘AI’ icon if the technology merely involved in the creation of a product; an ‘AI generated’ icon for those entirely produced by the technology with no human element involved except prompts; and an ‘AI modified’ icon for pre-existing, human-made content that has undergone manipulation with AI tools.How is the EU code different from regulations in India? The Ministry of Electronics and Information Technology, in February this year, notified an amendment to the IT Rules, 2021, that required users and social media platforms to label synthetically generated information (SGI) prominently for users to see. However, while the EU code requires developers of AI systems to include the label, India puts focus on social media, with the onus on users for the disclosure while relying on the platforms to enforce it. While the EU code seeks to bring about accountability from AI systems to improve responsible content generation, India has sought to regulate the final product with provisions for platforms to take down content that goes against the IT Rules. Also, the EU has laid down specific exemptions for creative content and put trust on human review, Indian rules are more generic and don’t provide such leeway.How have industry leaders responded to the new requirements? While most of the leading AI companies have signed on to the initiative, some have expressed reservations. While Google has also signed on to the code and noted that the initiative goes with its work to develop AI transparency tools, it is concerned about the regulatory complexity of the code. “If online content is flooded with overlapping AI labels and legal disclosures, it becomes harder for people to get the clear context they need”, Karen Massin, Google’s head of Government Affairs and Public Policy — EU Institutions, said in a statement. “Adding more regulatory complexity — as technical solutions are still evolving — could contradict Europe’s goals for competitiveness and simplification”, the statement said, adding that “it also risks confusing the people these rules are meant to help”. OpenAI has also welcomed the code and extended support to it, saying that it is “committed to doing our part by complying with the requirements that apply to our relevant products”. As AI tools become a part of how people build, imagine, and share, it’s important that people can understand where content comes from and interpret it with confidence, it said. However, the AI giant has acknowledged that the transparency push is still a nascent field as icons can be tampered with, and metadata can be removed. Saying that advancing the reliability of such techniques requires significant work among the AI ecosystem, it has called for the new code to be grounded in methods that are flexible enough to reflect limitations of current techniques. Published - July 29, 2026 10:32 am IST