Govt tightens rules on AI-generated content, cuts removal deadline to 3 hours from 36 & mandates labels
The Ministry of Electronics and Information Technology amended the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 to introduce stricter obligations on synthetically generated (AI) content
The deadline for intermediaries to remove unlawful content flagged via a court order or an authorised government notice was cut from 36 hours to roughly 3 hours
The amendment introduces a formal definition of "Synthetically Generated Information" (SGI) — content that appears reasonably authentic but is artificially or algorithmically generated, modified, or altered using a computer resource — and mandates visible/embedded labelling and provenance metadata for such content
Significant Social Media Intermediaries face the strictest obligations, including proactive detection, labelling, and blocking duties for synthetic audio-visual content such as deepfakes
IT Rules, 2021 — Intermediary Guidelines and Digital Media Ethics Code
The Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules were notified in 2021 under the Information Technology Act, 2000 to regulate social media intermediaries, digital news platforms, and OTT content. They replaced the 2011 Intermediary Guidelines and introduced due-diligence obligations intermediaries must follow to retain legal protection.
Key Details
- Framed under Section 87 (rule-making power) read with Section 79 of the IT Act, 2000
- Classify intermediaries into general "intermediaries" and "Significant Social Media Intermediaries" (SSMIs) — platforms with 50 lakh (5 million) or more registered users in India, which face enhanced obligations (grievance officer, compliance officer, nodal officer, monthly compliance reports, traceability of first originator for messaging apps)
- The 2026 amendment adds "Synthetically Generated Information" as a distinct regulated category, with mandatory labelling covering roughly 10% of the visual surface of an image or a stated portion of audio/video duration, plus embedded metadata for provenance
- Compliance timelines were tightened across the board — most notably the removal window for unlawful content (via court/government order) cut from 36 hours to about 3 hours
This amendment is a direct escalation of the 2021 Rules framework, adding a new, explicitly defined category (SGI/deepfakes) with faster takedown and mandatory labelling obligations layered onto the existing SSMI compliance architecture.
Section 79 of the IT Act — Safe Harbour and Its Conditions
Section 79 of the IT Act, 2000 is the "safe harbour" provision that shields intermediaries from liability for third-party content hosted on their platforms, provided they act only as passive conduits and observe prescribed due diligence.
Key Details
- Safe harbour protects an intermediary from liability for user-generated content as long as it does not initiate the transmission, select the receiver, or modify the content, and observes due diligence under the IT Rules
- If an intermediary fails to comply with the due-diligence obligations (including the new SGI labelling/takedown rules), it loses safe-harbour protection under Section 79(1) and becomes liable under the IT Act and other applicable laws, including the Indian Penal Code/Bharatiya Nyaya Sanhita
- Section 69A of the IT Act separately empowers the government to order blocking of content in the interest of sovereignty, security, public order, etc., following the procedure laid down in the Blocking Rules, 2009
- The Supreme Court's ruling in Shreya Singhal v. Union of India (2015) struck down Section 66A (criminalising "offensive" online speech) as unconstitutional but upheld Sections 69A and 79 with safeguards, requiring blocking/takedown orders to be reasoned and subject to due process
The 3-hour removal deadline and mandatory SGI labelling are new due-diligence conditions intermediaries must meet to retain Section 79 safe harbour — non-compliance exposes platforms to direct legal liability for deepfake and synthetic content circulated on them.
Deepfakes and Synthetic Media Governance
Deepfakes — AI-generated audio, video, or images that convincingly impersonate real people — have emerged as a governance concern globally due to risks of misinformation, fraud, non-consensual imagery, and election interference.
Key Details
- India does not have a standalone "deepfake law"; regulation has proceeded through amendments to the existing IT Rules, 2021 framework rather than new primary legislation
- The new rules require synthetic content to carry both a visible label/disclaimer and embedded metadata identifying it as machine-generated, aimed at making provenance traceable even after content is re-shared or edited
- Comparable international approaches include the EU AI Act's transparency obligations for AI-generated/deepfake content and mandatory labelling requirements emerging in other jurisdictions through 2026
- Enforcement responsibility sits with MeitY, with grievance redressal escalating through intermediary Grievance Officers, Grievance Appellate Committees (established under the 2021 Rules), and ultimately courts
This amendment situates India's deepfake response within the existing intermediary-liability model (rules + safe harbour conditions) rather than a fresh criminal statute, a distinction useful for comparing India's regulatory approach with other jurisdictions.
- Removal deadline for unlawful content flagged via court order/authorised government notice: cut from 36 hours to approximately 3 hours
- Significant Social Media Intermediary (SSMI) threshold: 50 lakh (5 million) registered users in India
- Governing framework: IT Rules, 2021, framed under Section 87 read with Section 79 of the IT Act, 2000
- New regulated category introduced: "Synthetically Generated Information" (SGI), covering deepfakes and AI-generated/altered audio-visual content
- Safe harbour for intermediaries under Section 79 is conditional on compliance with due-diligence obligations, including the new labelling and takedown requirements
- Landmark case on IT Act intermediary/speech provisions: Shreya Singhal v. Union of India (2015)