MeitY's October 8 Advisory Asks Platforms to Label 'Manipulated, Synthetically Generated or Misleading' Content -- Wider Than the Labelling Rules It Builds On

India's Ministry of Electronics and IT (MeitY) sent social media platforms an advisory dated October 8, 2026, issued by its Cyber Laws Division and signed by Joint Secretary Ajit Kumar. MediaNama first reported it, saying "Multiple sources have confirmed that the advisory, dated October 8, 2026, has been sent to platforms," and published its text. Deccan Chronicle and BestMediaInfo confirmed the core points.

What platforms are asked to do

• Label. "Identify and appropriately label/contextualise manipulated, synthetically generated or misleading content, particularly where such content is presented as depicting a current event."
• Enhanced diligence. BestMediaInfo lists the categories: "old or archived photographs and videos being passed off as current events, digitally altered or synthetically generated audio-visual content, impersonation of public authorities or individuals, and content likely to incite violence or other unlawful activity."
• Prioritise government complaints. Act on complaints "received from authorised Government agencies regarding content having potential implications for public order," and cooperate with law enforcement "within the timelines prescribed under the IT Rules, 2021."
• Process and users. "Review and strengthen" monitoring, escalation and grievance systems, and encourage users to "verify the authenticity, source and date of information before sharing or forwarding such content."

The advisory says the measures are meant to preserve public order "while respecting lawful expression and legitimate public discourse." It comes during protests in Delhi and other cities over the Election Commission's revision of electoral rolls, though BestMediaInfo notes "The document itself, however, does not refer to either the protests or the Election Commission."

How it relates to existing rules

MeitY amended the IT Rules on February 10, 2026, with the changes in force from February 20. MediaNama explains that their labelling duties apply to "synthetically generated information," which the Rules "confine to audio, visual or audio-visual content that appears real, so text-only content falls outside it." The new advisory's "labelling clause carries no such limit or refer to the February amendment anywhere." By adding "manipulated" and "misleading" content, MediaNama argues, it "leaves platforms to decide which posts are misleading without any fetters or checks."

Other gaps: the advisory asks for action "with immediate effect" but names no specific consequence; BestMediaInfo says it "does not cite a specific legal provision for the directions or spell out penalties for non-compliance." It does not define "enhanced diligence," which MediaNama notes differs from the "due diligence" the IT Rules require, or say what "priority" means for a platform's review queue. Its clause numbering jumps from (iv) to (vi).

What it means for AI products

For platforms, this is a moderation workload question. Classifying "misleading" content at speed means automated detection plus human review, and an advisory with no definitions pushes the hardest judgement calls onto trust-and-safety teams. For generative AI companies serving India, the message is that labelling expectations are moving faster than the notified rules. The day before, IT Minister Ashwini Vaishnaw said a consultation paper on AI regulation, with deepfakes among its focus areas, would arrive within a month, as this digest reported on October 9.

What to watch. Whether platforms read the labelling ask as covering text as well as media; whether the coming consultation paper writes "manipulated" and "misleading" into binding rules; and how "priority" for government complaints is put into practice.

Analysis: companies shipping image, video or voice generation in India should make output provenance (embedded watermarks and metadata) a default, not a feature, so platform labelling can rely on a signal rather than guesswork.