AI content now under legal framework as Govt tightens IT rules

NEW DELHI: The Central Government has amended the IT Rules to curb fraud, fake propaganda and other crimes on social media using Artificial Intelligence (AI). The new rules will come into effect on February 20. For the first time, AI-generated content has been brought under a legal framework. It will be classified as “Synthetically Generated Information” (SGI).

Under the new rules, social media platforms must clearly label AI-generated content. Users should be able to easily understand that the content is created using AI and identify its origin. The label cannot be removed. This will help in identifying deepfake videos, artificially created audio clips and other synthetic content.

The IT Ministry made the changes using its powers under Section 87 of the IT Act. The amended rules are titled the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026.

If the Centre directs that certain posts of an urgent nature be removed, platforms must take them down within three hours. Earlier, they had 36 hours to do so.

Content related to child sexual abuse, obscene material, fake voice messages, explosives, and deepfake videos that mislead people will be treated as criminal offences under laws such as the Bharatiya Nyaya Sanhita, the POCSO Act and the Explosives Act.

Synthetically Generated Information
“Synthetically Generated Information” refers to audio, visual or other content that is artificially created to appear real and authentic. However, routine editing such as colour correction or noise reduction that does not change the original content will not fall under these rules.

Social media

1. The new regulations impose stricter controls on social media platforms such as Facebook, Instagram and YouTube.
2. Before allowing uploads, platforms must ask users whether the content is AI-generated.
3. They must also use automated tools to cross-check
4. Platforms are required to examine the format, source and nature of the content.
5. They must issue warnings to users every three months about the penalties for misuse of AI-generated content.

No watermark needed
The earlier 2025 draft proposal that required watermarking of AI content has been dropped. The draft had suggested reserving 10 percent of screen space for a watermark. Social media companies had informed the government about practical difficulties in implementing this requirement.