The Ministry of Electronics and Information Technology (MeitY) is set to withdraw the requirement to make large, fixed-size watermarks mandatory on AI-generated content, two officials familiar with the matter said, as the government prepares to notify final amendments to the Information Technology (Mediation Guidelines and Digital Media Ethics Code, this week, 221 Rules).
Officials said the government is moving away from setting a 10% threshold.Draft amendments published in October 2025 proposed that intermediaries offering tools to create or modify artificially generated information must embed distinctive labels or identifiers. For visual content, these labels were proposed to cover at least 10% of the surface display area, while for audio content, an audible marker was proposed for the first 10% of its duration.
For platforms like Instagram or YouTube, this means large, always-visible watermarks or audio tags on every AI-generated post or video, covering parts of the content and influencing the viewing experience.
Only labels that occupy at least 10% of the screen space are being excluded. Content must still carry a clear indication that it is AI-generated.
Officials said the ministry is moving away from setting a 10% threshold in the final version. “If I watch a video, I should know that something is AI-generated,” one of them said.
A second official said the industry opposed the amendment. “We have considered their request.”
The change followed pushback from industry groups who argued that the rule was too strict and technically difficult to apply across formats and devices. The Internet and Mobile Association of India (IAMAI), whose members include Google, Meta, Snapchat, WhatsApp, Zomato, Groww, Airbnb, Amazon, Apple, Jio, Airtel, Netflix, Ola and Reddit, consulted the companies and submitted a representation to the ministry outlining their concerns.
In the submission, IAMAI said the labeling requirement is overly prescriptive and may disrupt the user experience, particularly for audio and video content. It noted that visible watermarks can be easily cropped or altered, metadata can be stripped, and there is no common industry standard to ensure consistent implementation. The company also identified privacy, interoperability and compliance burdens, particularly for smaller companies.
More Read
Industry representatives said their biggest concern remains a separate provision, Rule 4(1A), which will require key social media intermediaries such as Instagram, YouTube, Facebook, etc. to receive user declarations on whether content is synthetically generated, put in place “reasonable and proportionate” technical measures to verify those declarations, and ensure that such content is clearly labelled. Organizations have argued that verifying declarations at scale can be difficult in practice.
The ministry, after releasing the draft rules in October 2025, had invited public feedback till November 6, which was extended to November 13 after industry requests.
