India directs social media platforms to remove illegal content within three hours

India has introduced new rules requiring social media companies to remove unlawful content within three hours of notification, a sharp reduction from the previous 36-hour window.

The updated guidelines, effective from 20 February, apply to major platforms including Meta, YouTube, and X, and cover AI-generated content as well. The government has not explained why the takedown timeframe was shortened.

Critics warn the move could intensify government oversight of online content and potentially lead to censorship in the world’s largest democracy, home to over a billion internet users. In recent years, Indian authorities have used Information Technology rules to direct social media firms to remove material deemed illegal under national security and public order laws. Transparency reports indicate that more than 28,000 URLs were blocked following government requests in 2024.

The amendments also introduce rules for AI-generated material, defining it as audio, video, or other content created or altered to appear real, such as deepfakes. Ordinary editing, accessibility features, and legitimate educational or design work are excluded.

Platforms must label AI-generated content clearly and, where possible, add permanent markers to trace its origin. These labels cannot be removed, and automated tools must be used to detect and prevent illegal AI content, including deceptive or non-consensual material, false documents, child sexual abuse content, explosives-related material, and impersonation.

Digital rights groups and technology experts have raised concerns about the feasibility and implications of the new rules. The Internet Freedom Foundation warned that the three-hour window would turn platforms into “rapid-fire censors,” leaving little time for human review and forcing over-reliance on automation.

Anushka Jain, a research associate at the Digital Futures Lab, noted that while the AI labelling requirement could improve transparency, the tight deadline risks pushing platforms toward fully automated moderation, increasing the likelihood of wrongful removal. Delhi-based technology analyst Prasanto K Roy described the regulations as “perhaps the most extreme takedown regime in any democracy,” highlighting the difficulty of compliance without extensive automation and minimal human oversight.

On AI labelling, Roy added that while the intent is positive, tamper-proof and reliable technologies are still under development.