Origin Labels Become Mandatory Immediately
The Government of India has notified a new chapter of the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, targeting the sharp rise in deepfake incidents. Branded informally as the “IT Amendment Rules 2026”, the policy requires that any AI-generated photo, video, or audio clip carry a persistent origin label or watermark that covers at least 10% of the visible frame.
Union IT Minister Ashwini Vaishnaw told reporters that the watermark specification will be aligned with Bureau of Indian Standards (BIS) guidelines so platforms, creators, and broadcasters adopt a uniform cue. “People should not have to guess whether a clip is synthetic. The label must be as obvious as the content itself,” he said, adding that non-compliance can trigger penal action under the IT Act.
36-Hour Removal Clock For Platforms
Besides labeling, the amendment shortens the turnaround time for takedowns. Social networks, chat apps, and short-video platforms now have 36 hours to remove or disable any deepfake flagged by a user, court, or government notice. Missing the deadline could strip intermediaries of “safe harbor” protection under Section 79, exposing them to civil or criminal liability for the hosted content.
Cyberlaw expert Pavan Duggal called the timeline demanding but necessary. “Election integrity and financial scams are on the line. Platforms already have war rooms for terrorism and CSAM takedowns, so deepfake workflows must receive the same priority,” he said. Several companies are reportedly reconfiguring trust-and-safety dashboards to highlight flagged synthetic media faster.
Verification API For Newsrooms
To help fact-checkers, MeitY’s Indian Computer Emergency Response Team (CERT-In) is building a watermark verification API. Newsrooms, election authorities, and law-enforcement units will be able to upload a suspicious clip and immediately check whether the label is genuine. The beta is expected before the 2026 state election calendar. Officials said exemptions for satire, archival footage, and art projects will be finalized after a 30-day consultation, but political advertising and financial advice will remain within mandatory labeling zones.
What’s Next
The ministry is working with startups at IIIT-Hyderabad to design tamper-resistant watermarks backed by distributed ledgers. Quarterly compliance reports are now compulsory, and repeat offenders can face temporary blocking orders. For creators, the guidance is straightforward: document which model, prompt, and editing tools were used, embed an origin label before uploading, and retain logs in case a takedown query arrives.
With these changes, India joins the EU and Singapore in drafting some of the most prescriptive regulations on synthetic media, signaling that 2026 will be a year of zero tolerance for unlabeled AI content.
