Abstract
The February 2026 amendments to India’s intermediary rules introduced a defined category of synthetically generated information (SGI), reduced the deadline for acting on specified court orders or authorised government notices to three hours, and set a separate two-hour period for certain complaints about intimate or impersonating content. This article places those changes within Section 79 of the Information Technology Act and the actual-knowledge standard in Shreya Singhal. It asks whether a short clock for removal can protect victims while preserving careful review of lawful speech, and proposes proportionate safeguards.
I. Introduction
Synthetic images, cloned voices and impersonating videos can spread quickly and cause reputational, financial and personal harm. The Ministry of Electronics and Information Technology notified amendments to the intermediary rules on 10 February 2026, effective 20 February 2026. The change that attracted the most attention is the three-hour period in Rule 3(1)(d) for acting after a court order or a reasoned written notice from an authorised government officer concerning unlawful information. A separate two-hour period applies to the complaints described in Rule 3(2)(b), including specified private, intimate and impersonating material. The distinct triggers and categories matter when assessing the rule’s effect on expression.
This article first explains Section 79 and the actual-knowledge standard in Shreya Singhal v. Union of India. It then considers the court orders that preceded the amendment, describes the 2026 rules and tests their practical effect on lawful speech. A comparison with European and Singaporean approaches informs the proposed safeguards.
II. The Architecture of Intermediary Liability Before 2026
A. Section 79 and Conditional Immunity
Section 79 of the Information Technology Act, 2000 grants an intermediary conditional immunity from liability for third-party content, provided the intermediary observes due diligence and, upon receiving actual knowledge of unlawful content, expeditiously removes or disables access to it. The provision is the statutory foundation of every subsequent takedown regime, including the 2026 Amendment: it is what an intermediary stands to lose if it fails to comply with a takedown order in time.
B. Shreya Singhal and the Actual-Knowledge Standard
In Shreya Singhal, the Supreme Court read down Section 79(3)(b): actual knowledge for this purpose requires a court order or an appropriate government notification, rather than a private allegation alone. That interpretation reduces the pressure on intermediaries to remove lawful content merely because a complaint has been made. It remains the starting point for assessing the shorter compliance period.
C. The IT Rules, 2021 and the Thirty-Six-Hour Baseline
Before the 2026 amendments, Rule 3(1)(d) generally allowed thirty-six hours to act on a qualifying order or notice. The rules also provided a grievance process and intermediary due-diligence obligations. They did not define SGI; synthetic impersonation was addressed through more general restrictions, including those concerning privacy and unlawful content.
III. Filling the Gap: Courts as First Responders
Before the 2026 amendment, courts often addressed synthetic impersonation through interim orders protecting personality and publicity interests. In Arijit Singh v. Codible Ventures LLP, the Bombay High Court considered unauthorised use of a performer’s voice and likeness in AI-generated content. Such cases show how a person may seek targeted relief after identifying the offending material.
Other disputes involving impersonation and fraudulent promotion similarly prompted requests to remove particular posts and identify those responsible. These orders helped the affected claimants, but relief depended on litigation and the facts of each case.
The case-specific approach also required courts to distinguish impersonation from other speech that might merely be alleged to be defamatory. That distinction matters when designing a general rule for intermediaries.
These cases illustrate both the value and the limits of injunctions. Litigation can identify a concrete wrong and provide focused relief, but it usually follows publication and may be difficult for a victim to pursue. The 2026 rules attempt a more general response; their breadth makes clear safeguards especially important.
IV. The 2026 Amendment: Structure and Stated Rationale
A. Synthetically Generated Information
Rule 2(1)(wa) defines synthetically generated information as audio, visual or audio-visual information artificially created or altered using a computer resource so that it appears real or authentic and portrays a person or event as indistinguishable from a natural person or real event. The definition focuses on the perceived result of the content, rather than naming particular software. Its express exclusions for routine editing and other non-deceptive processing also matter when assessing whether a particular item falls within the rule.
B. Duties Concerning Synthetic Content
Rule 3(3) imposes additional due-diligence duties concerning SGI on intermediaries that enable the creation, generation, modification or alteration of such information. It calls for reasonable and appropriate technical measures against specified unlawful material and for the labelling of lawful SGI. These duties must be distinguished from the separate notice-based removal obligation in Rule 3(1)(d), which concerns unlawful information more broadly.
C. Labelling and Provenance
The notified framework requires intermediaries covered by Rule 3(3) to label lawful SGI prominently and, where technically feasible, to embed provenance information. The final text does not impose the fixed ten-per-cent visual or audio label proposed in an earlier draft. Significant social-media intermediaries also have additional declaration and verification duties under Rule 4(1A). These requirements differ from the separate first-originator provision applicable to certain messaging services; the amendment does not create a universal first-originator record for every item of SGI.
D. The Three-Hour Mandate
Rule 3(1)(d) now requires an intermediary to remove or disable access to unlawful information within three hours after receiving a court order or a reasoned written intimation from an authorised government officer. Rule 3(2)(b) separately requires action within two hours on specified complaints involving private areas, nudity, sexual content or impersonation, including certain morphed images. The three-hour rule is not confined to synthetic content, and the two-hour rule has a different complaint trigger. Failure to observe due diligence may put Section 79 safe-harbour protection at risk under Rule 7, subject to the applicable law and facts; the Rules do not make every missed deadline an automatic finding of publisher liability.
V. Testing the Regime
A. Operational Feasibility
A three-hour deadline presupposes a capacity to receive, authenticate and act on qualifying orders at all hours. That is more achievable for a large platform with specialist staff than for a small intermediary. Automated detection can help prioritise reports, but it can also misclassify satire or edited documentary material. The practical question is whether the process allows enough human review to distinguish an order’s scope, the precise content identified and the action legally required.
B. Constitutional Proportionality
The doctrinal concern is the practical effect of a short compliance clock. Shreya Singhal narrowed actual knowledge for Section 79 purposes to a court order or appropriate government notification, protecting lawful expression from removal solely on a private allegation. Even where that threshold is met, an intermediary may have little time to resolve an ambiguous notice before acting. The risk of over-removal should therefore be considered alongside the legitimate aim of preventing fast-moving harm, with attention to the clarity of notices, opportunities for review and the different circumstances of platforms.
C. Over-Removal and the Chilling Effect on Lawful Speech
The short deadline can encourage broad removal where a notice is unclear or content depends on context, such as satire, journalism or artistic expression. Technical tools can identify likely matches, but cannot always decide whether a particular communication is unlawful. A clear written notice identifying the legal ground and specific material, as Rule 3(1)(d) requires for government intimations, is an essential safeguard against such errors.
D. Cross-Border Enforcement Gaps
Content and platforms may cross borders, complicating investigation and enforcement. A three-hour deadline does not by itself ensure that a foreign service will preserve evidence, identify an uploader or prevent the material from circulating elsewhere. Clear, specific notices and cooperation under the applicable legal process remain important.
VI. A Comparative Perspective
The international comparison is useful but limited. The European Union’s Terrorist Content Online Regulation sets a one-hour removal deadline after a competent authority’s order for terrorist content. Its subject matter and procedural framework differ from India’s broader Rule 3(1)(d). The EU Digital Services Act uses notice-and-action processes and additional obligations for its largest platforms. India’s three-hour period should therefore be evaluated against its own breadth, decision process and appeal mechanisms, rather than described as the world’s shortest deadline.
VII. Towards a Calibrated Regime
The harm targeted by the rules is real, particularly where intimate or impersonating material spreads quickly. The existing two-hour complaint route and three-hour order route already recognise different triggers. Further refinement could improve accuracy while retaining prompt action.
First, the government could publish clearer guidance on the required specificity of notices and on preserving a record of the decision. Second, an expedited review route could restore wrongly removed time-sensitive speech. Third, compliance guidance could address the resources of smaller intermediaries without diluting the statutory duty to act. Fourth, consistent, visible SGI labelling would help users understand content without treating every synthetic work as unlawful.
VIII. Conclusion
The 2026 amendments respond to serious harms from synthetic content and other unlawful information. Their three-hour order deadline, two-hour complaint route and SGI duties have different legal triggers and should be assessed separately. Clear notices, prompt review of disputed removals and workable guidance for intermediaries would help preserve both victim protection and lawful expression.
Primary sources
MeitY: consolidated intermediary rules as amended in February 2026