• The Union Government notified the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules, 2026 on February 10, 2026.
• The amendment specifically targets the rising threat of deepfakes and misinformation by mandating a swift 3-hour takedown window and compulsory labelling for AI-generated content.

About IT Rules Amendment 2026:

• The amendment is a revision of the existing IT Rules, 2021, aimed at bringing Synthetically Generated Information (SGI) under strict legal oversight. It defines SGI as any audio, visual, or audio-visual content created or altered algorithmically to appear real or indistinguishable from a natural person or real-world event.

Key Amendments to the IT Rules:

• Legal Recognition of “Synthetically Generated Information”: For the first time, Indian law provides a clear, technical definition of Synthetic Content. It covers any media (audio, video, or images) created or modified by algorithms to look or sound authentic.
• The Focus: It specifically targets deepfakes and AI impersonation.
• The Nuance: It smartly carves out exceptions for good faith editing—like basic filters, accessibility features (text-to-speech for the blind), or academic research—so that innovation isn’t accidentally criminalized.
• Mandatory Labelling & Metadata: Transparency is no longer optional; it must be baked into the file itself.
• Visual/Audio Labels: If a video is AI-generated, it must have a visible watermark. If it’s audio, it must start with a spoken disclaimer.
• Digital Fingerprints: Platforms must embed provenance markers (metadata) that stay with the file even if it’s shared elsewhere. This allows investigators to trace the origin of a deepfake back to the specific AI tool used to create it.
• Prohibition on Illegal AI Content: The government has moved from reactive to preventative. Intermediaries (like Instagram or X) must now use automated AI filters to block the upload of:
• CSAM & NCII: Child abuse material and non-consensual revenge porn deepfakes.
• Public Safety Risks: AI-generated instructions on how to build explosives or illegal weapons.
• Deception: Content designed to impersonate high-ranking officials or create false electronic records to commit fraud.
• User Declaration Mechanism: The burden of honesty is placed on the user.
• Self-Disclosure: When you upload content to a major platform, you must declare if it was made with AI.
• Verification: Platforms cannot just take your word for it; they are legally required to use their own technical tools to double-check if your real video is actually a deepfake.
• Shortened Takedown Timelines: In the age of viral content, 36 hours was considered an eternity.

The new rules require warp-speed action:

• 3 Hours: For content deemed illegal by a court or the government.
• 2 Hours: For the most sensitive violations, like non-consensual deepfake nudity, where every minute of exposure causes trauma.
• Faster Grievances: Platforms must acknowledge your complaint in 7 days (down from 15), forcing them to staff up their moderation teams.
• Law Enforcement Coordination: This amendment bridges the gap between digital policy and criminal justice.
• BNS Integration: References are now aligned with the Bharatiya Nyaya Sanhita (BNS), 2023, replacing the old IPC.
• Identity Disclosure: If a crime is committed via AI, platforms must reveal the creator’s identity to the police. This is designed to end the anonymity shield that often protects deepfake creators.
• Safe Harbour Clarification: This is the Carrot and Stick approach for Big Tech.
• Section 79 Protection: Platforms generally aren’t sued for what users post (Safe Harbour).
• The Condition: Under the 2026 rules, if a platform fails to label AI or misses a 3-hour takedown window, they lose this protection. They can then be sued as if they were the ones who created the illegal content.

Significance of the Amendment:

• Combating Viral Misinformation: Prevents the rapid spread of fake news before it can cause real-world damage.
• E.g. In the 2025 state elections, fabricated videos of candidates making inflammatory speeches were quelled by rapid intervention from the Cyber Crime Coordination Centre.

• Protecting Individual Dignity: Provides a fast-track remedy for victims of non-consensual AI-generated imagery.
• E.g. The 2024 Rashmika Mandanna deepfake incident highlighted the need for near-instant removal to prevent irreversible reputational harm.

• Ensuring Electoral Integrity: Guards against the use of AI influencers or morphed videos to sway voters during the sensitive Model Code of Conduct period.
• E.g. During recent local polls, AI-cloned voices of deceased political leaders were used for campaigning, prompting calls for strict disclosure.

• Strengthening Business Accountability: Forces global tech giants to invest in India-specific moderation and detection technologies.
• E.g. Companies like Meta and X had to expand their Indian Grievance Officer teams in late 2025 to meet the increasingly tight response windows.

• Alignment with New Criminal Laws: The rules replace references to the IPC with the Bharatiya Nyaya Sanhita (BNS), 2023, streamlining the legal process.
• E.g. Police in Delhi recently used BNS provisions alongside IT rules to fast-track a case involving AI-based financial digital arrest scams.

Challenges Associated with Implementation:

• Technical Accuracy of Detection: Automated tools often struggle to differentiate between high-quality deepfakes and actual real footage.
• E.g. In 2025, several genuine satire videos were accidentally flagged and removed by AI filters on major video-sharing platforms.

• Resource Constraints for 3-Hour Takedowns: Smaller intermediaries may find it impossible to maintain 24/7 legal teams capable of acting within 180 minutes.
• E.g. Regional social media apps in South India reported difficulty in 2024 meeting the high costs of compliance for real-time content moderation.

• Potential for Censorship by Proxy: Fears that the government might use the short window to suppress legitimate dissent or political parody.
• E.g. Ongoing litigation in the Karnataka High Court (X Corp v. Union of India) debates whether takedown powers are being used too broadly.

• Traceability vs. Privacy: Embedding metadata and provenance markers could potentially compromise the end-to-end encryption of messaging apps.
• E.g. In 2025, privacy advocates argued that the new metadata rules could act as a “backdoor” to identifying anonymous whistle blowers.

• Complexity of Blended Content: Difficulty in regulating videos that are 90% real but contain small, crucial AI-altered details.
• E.g. A 2024 controversy involved a real protest video where the audio was subtitled with AI-generated translated hate speech that didn’t exist.

Way Ahead:

• Standardized Watermarking: Developing a global industry standard for invisible digital watermarks that survive compression and re-uploads.
• Capacity Building for Law Enforcement: Training local police units (beyond just the DIG rank) to accurately identify and report synthetic harms.
• Public Awareness Campaigns: Educating citizens on how to spot telltale signs of deepfakes, reducing the reliance on takedowns alone.
• Independent Review Mechanism: Establishing an autonomous body of experts to review takedown orders to prevent political misuse.
• Incentivizing Research: Providing grants to Indian startups building advanced AI-detection tools specifically for regional Indian languages.

Conclusion:

• The 2026 IT Rules amendment marks a decisive end to the wild west era of unregulated generative AI in India by shifting the burden of truth onto platforms. While the 3-hour takedown window poses a massive logistical challenge, it reflects the government’s priority of safety over safe harbour. Success will depend on balancing these strict enforcement measures with the protection of free speech and user privacy.