India Proposes First Major AI Regulation Targeting Social Media Content

The Indian government has taken a significant step toward implementing the country’s first substantial artificial intelligence regulation, focusing specifically on social media platforms and their users. On October 22, the Ministry for Electronics and Information Technology released draft amendments to the Information Technology Rules 2021 that would govern AI-generated and synthetic content circulating on social media networks.

As generative AI technology becomes increasingly accessible, allowing anyone to instantly create text, images, music, and videos, governments worldwide are grappling with how to mitigate associated risks, particularly the spread of misinformation. Despite India’s generally optimistic stance toward AI, evidenced by its upcoming AI Impact Summit 2026, these proposed regulations appear to follow the more restrictive digital governance models implemented in China and the European Union.

China recently introduced rules requiring AI-generated content to be clearly labeled to distinguish it from human-created material, while the EU has mandated that deepfakes must be marked as such. Following this international trend, India’s proposed regulations would require social media companies and services that enable AI content creation or modification to clearly label “synthetically generated information.”

Under the draft rules, platforms would need to embed metadata in synthetic content to ensure users can identify it, superimpose labels covering at least 10 percent of any visible content, and verify declarations by users who share synthetic content. These requirements would be part of the expanded due diligence obligations for social media companies under the IT Rules.

Perhaps most significantly, the draft amendments would allow social media companies to remove content based solely on user complaints, without requiring a court or government order. This represents a substantial shift in India’s digital content governance framework.

Currently, India’s IT Act provides “safe harbor” protection to social media platforms, granting them immunity from prosecution for user-generated content. This legal shield has been crucial for allowing these platforms to innovate and grow without being held legally responsible for the vast amounts of diverse content they host. Without such protection, social media likely would not function as the participatory space users value today.

The proposed rules would weaken this safe harbor regime in two crucial ways. First, by allowing platforms to take down content based merely on user complaints, social media companies would step beyond their role as neutral conduits of information. Second, larger platforms would need to continuously monitor content, as merely being “aware” of synthetic content’s existence – whether detected voluntarily or through complaints – would trigger a loss of legal immunity.

Legal experts note that this approach contradicts the “actual knowledge” standard established by the Supreme Court in the landmark case of Shreya Singhal v. Union of India, which shifted the responsibility for determining content illegality from intermediaries to courts and the government. By reversing this principle, the draft rules would force platforms to become arbiters of online speech.

This outcome runs counter to recent parliamentary sentiment, as two parliamentary standing committees have expressed bipartisan concern about the powers already held by social media companies, suggesting they need more accountability rather than expanded content moderation powers.

The labeling mandate itself presents practical challenges. While labeling helps users identify AI-generated material, experts caution that implementing such requirements at scale is difficult. Social media platforms will inevitably miss some synthetic content, which could inadvertently enhance the perceived credibility of unlabeled material. Conversely, human-produced content might be incorrectly labeled as synthetic, undermining trust in legitimate information.

A more effective approach might focus on standardizing labels and improving user literacy. Organizations like the Coalition for Content Provenance and Authenticity are already developing open technical standards to establish content origins. Encouraging broader adoption of such industry standards could ensure more consistent labeling while maintaining flexibility for continued innovation.

Studies show that while AI content labels may increase transparency, they don’t significantly change how persuasive that content is to users. As our information landscape becomes increasingly artificial, user vigilance and media literacy become increasingly important complementary safeguards.

Ultimately, addressing synthetic content challenges requires a collaborative approach between government and industry, grounded in established principles and supported by shared resources, rather than placing the burden primarily on either platforms or users.