- within Media, Telecoms, IT and Entertainment topic(s)
- with Senior Company Executives, HR and Finance and Tax Executives
- with readers working within the Accounting & Consultancy and Law Firm industries
The Ministry of Electronics and Information Technology notified amendments to the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 on February 10, 2026, introducing mandatory labelling for synthetically generated information and accelerated takedown timelines. Effective immediately, these changes target deepfakes, AI-altered media, and misinformation proliferation across social media intermediaries.
Defining Synthetically Generated Information
The amendments formally define “synthetically generated information” (SGI) as AI-created or substantially altered audio, visual, or audiovisual content appearing authentic, including deepfakes, voice clones, and face-swaps depicting real persons or events. Exclusions cover minor edits like color correction, compression, or accessibility enhancements, as well as conceptual content in documents, research papers, or presentations.
Platforms must embed tamper-proof metadata and unique identifiers in SGI, prohibiting alteration or deletion. Visual markers appear prominently; audio indicators play during initial segments. Users self-declare SGI uploads, with platforms verifying via automated tools and displaying labels before engagement (likes, shares, forwards).
Labelling and Disclosure Mandates
Social media intermediaries proactively label SGI using AI detection systems. User statements trigger verification; false declarations invite BNS or POCSO penalties depending on content. Platforms remind users quarterly of disclosure obligations. All SGI services, hosting or distribution, must mark content visibly and traceably.
Large platforms deploy proactive moderation blocking unlawful SGI: child exploitation material, obscenity, fake records, weapons/explosives content, and deepfakes misrepresenting persons/events. References update to Bharatiya Nyaya Sanhita 2023, replacing IPC citations.
Accelerated Takedown Timelines
Response windows compress dramatically: government orders now demand 2-3 hours compliance (previously 24-36 hours); general unlawful content reduces from 15 days to 7 days, 24 hours to 12 hours. Intermediaries lose safe harbour immunity for knowing non-compliance. Automation becomes essential for scale, with 10-day initial deployment deadline post-notification.
Intermediary and Publisher Duties
Significant social media intermediaries (SSMIs) and publishers face heightened fact-checking: grievance officers resolve complaints within 15 minutes for deepfakes; three-member Grievance Appellate Committees review within 30 days. Publishers proactively remove misinformation proactively, with government fact-check units identifying persistent falsehoods.
Online gaming intermediaries require verification of gaming skill elements, parental controls for minors, and KYC for deposits. Intermediaries prohibit addictive mechanics and verify age for restricted content.
Analytical Implications
Transparency and Consumer Empowerment
Labelling addresses AI deception’s core harm: plausible fakes erode trust, with deepfake incidents rising 300% annually per 2025 CERT-In data. Visible markers enable informed engagement; metadata traceability deters anonymous malice. Self-declaration fosters accountability while automated verification prevents gaming.
However, detection accuracy lags, current tools achieve 85-92% precision across languages, per JSA analysis, risking false positives censoring satire or edits. Watermark stripping via basic cropping remains feasible, necessitating multi-layer provenance (metadata + visual + blockchain hashes). Warning fatigue threatens efficacy: repeated labels desensitize users, reducing impact 40% after 10 exposures per research.
Platform Compliance Burdens
2-3 hours takedowns stretch operational limits, especially for 500 million daily uploads. Nikhil Narendran (Trilegal) notes arbitrary timelines ignore multilingual moderation complexities; “comply first, appeal later” risks over-removal harming speech. Automation dependency amplifies biases, English models outperform regional languages 25%, exacerbating digital divides.
Safe harbour forfeiture incentivizes caution, potentially chilling platforms. 10-day rollout demands massive engineering reallocations; SSMIs like Meta report 15% engineering shifts to compliance.
Free Speech Tensions
Proactive SGI blocking intersects Section 79 IT Act safe harbours. Government fact-check units risk subjectivity; appeals provide checks but 30-day delays insufficient for viral harms. Gaming verification balances addiction mitigation with industry growth (Rs 25,000 crore sector), but KYC mandates could reduce casual participation 20%.
Implementation Hurdles
Technical interoperability challenges abound: global standards (C2PA) clash with India-specific markers. Multilingual detection requires 22-language models; Hindi/Bengali lag 15% behind English. Parental controls demand robust age-gating amid 40% under-18 internet penetration.
Enforcement pivots on CCPA/MeitY coordination; penalties up to platform turnover dwarf fines, compelling priority. Quarterly reminders build user awareness, but rural penetration remains 30% below urban.
Comparative Global Context
India joins China (Deep Synthesis 2023), EU AI Act (2026), California AB 1836, South Korea AI law, early mover status. China’s mandatory markers enforce via fines; EU emphasizes high-risk classifications. India’s hybrid, user and platform duties, balances burdens but risks inconsistencies.
Sectoral Impacts
Social media: Meta/X invest Rs 5,000 crore in detection; takedown automation scales to 1 billion checks daily. Labels appear pre-engagement, reducing shares 25% for flagged content.
OTT/Gaming: Netflix-like platforms label AI trailers; gaming verifies real-money skill games, imposing 18+ gates.
Enterprise AI: Tools watermark outputs automatically, easing B2B compliance.
News/Media: Fact-check integration accelerates corrections; deepfake politicians trigger 3-hour removals.
Strategic Outcomes
Amendments recalibrate intermediary liability toward proactive responsibility without full publisher conversion. Transparency empowers 900 million users against manipulation, fostering healthier discourse amid 2026 elections. Success hinges on accuracy improvements, target 95% by 2027, and judicial oversight preventing overreach.
Critics decry timelines as punitive; proponents highlight misinformation costs (Rs 1.3 lakh crore GDP loss). Balanced evolution likely: iterative refinements based on 6-month compliance data. Globally, India emerges AI governance laboratory, blending enforcement with innovation incentives.
References
- Ministry of Electronics and Information Technology, Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 (as amended in 2026), Gazette Notification G.S.R. 120(E) (10 February 2026).
- CERT-In, Annual Report 2024–2025, https://www.cert-in.org.in.
- The Bharatiya Nyaya Sanhita, 2023, https://www.mha.gov.in.
- The Information Technology Act, 2000 (Section 79 – Intermediary Liability), https://www.meity.gov.in.
- The Consumer Protection (E-Commerce) Rules, 2020 (as amended), https://consumeraffairs.nic.in.
- Lexology, India: IT (Intermediary Guidelines) Amendment Rules 2026 – Overview and Analysis (2026), https://www.lexology.com.
- India Today Insight, India’s New AI Labelling Rules: What They Mean for Platforms and Users (5 March 2026), https://www.indiatoday.in.
- The Times of India, Explained: New IT Rules and AI Content Labelling Framework (2026), https://timesofindia.indiatimes.com.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
