India Mandates Watermarks for AI-Generated Content
India has enacted its first legislation targeting digital media ethics, which mandates watermarks for any "significantly modified or AI-generated content," including photos and videos. The new rules are driving discussion in global photography and design communities about content provenance, creative freedom, and the need for transparent attribution in AI-assisted workflows.
- The new rules are an amendment to the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules of 2021, and they came into effect on February 20, 2026. The Ministry of Electronics and Information Technology (MeitY) is the governing body that notified these changes. - The regulations are not just for social media platforms; any intermediary that facilitates the creation or sharing of synthetically generated information must label it clearly and prominently on the content itself. They are also required to embed persistent metadata or unique identifiers to the extent that it is technically feasible. - Significant social media intermediaries are required to have users declare whether their content is synthetically generated before it is uploaded. These platforms must then use automated tools to verify these declarations. - The initial draft of the rules included a controversial proposal for watermarks to cover at least 10% of AI-generated visuals, but this was dropped from the final version after industry bodies, including Google, Meta, and Amazon, raised concerns about its rigidity and implementation challenges. - These amendments also significantly shorten the timeframes for content removal; platforms now have as little as three hours to take down certain types of unlawful content, a drastic reduction from the previous 36-hour window. Non-consensual intimate imagery, including AI-generated deepfakes, must be taken down within two hours. - Globally, other major regulatory bodies are also implementing measures for AI content transparency. The European Union's AI Act, for instance, requires that providers of generative AI systems ensure that the outputs are marked in a machine-readable format and are detectable as artificially generated or manipulated. - China has also implemented regulations that require AI-generated content to be marked. Their rules mandate an "explicit watermark" that is visible and does not interfere with the user's experience. - The technical robustness of watermarking is a key area of discussion, as research has shown that watermarks can sometimes be removed or even reverse-engineered, potentially allowing bad actors to disguise AI-generated text as human-written or vice-versa.
