India Mandates AI Media Labeling
India has amended its IT Rules, effective February 20, to require the labeling of all AI-generated media and to shorten content takedown response times to 2-3 hours. The new rules require platforms to maintain detailed operating procedures and evidence kits. The Ministry of Electronics and Information Technology framed the measures as essential for promoting "safe and trusted AI innovation" in India.
- The primary penalty for non-compliance is the loss of "safe harbor" protection under Section 79 of India's Information Technology Act. This exposes platforms to criminal liability for user-generated content under other laws, including the Bharatiya Nyaya Sanhita (the new penal code) and the Protection of Children from Sexual Offences (POCSO) Act. - The accelerated takedown timeline mandates removal of content flagged by a court or government agency within three hours, a significant reduction from the previous 36-hour window. For non-consensual intimate imagery, the deadline is even shorter, at two hours. - These amendments to the IT Rules of 2021 were prompted by a series of viral deepfake incidents, including fabricated videos of Bollywood actors and synthetic voice clones used in political campaigns, which raised alarms about the potential for widespread misinformation. - In addition to labeling, platforms must embed persistent metadata or unique identifiers in AI-generated content to help trace its origin. Intermediaries are prohibited from allowing the removal or suppression of these labels or identifiers. - During consultations, tech industry bodies pushed back against initial proposals, leading the government to soften a requirement for labels to cover at least 10% of an image to the final rule of being "prominently" displayed. - Digital rights organizations, such as the Internet Freedom Foundation, have raised concerns that the extremely short compliance window could compel platforms to rely heavily on automated content moderation, increasing the risk of erroneous takedowns and censorship. - The rules require significant social media intermediaries to not only ask users to declare if their content is AI-generated but also to use technical measures to verify these declarations. - Platforms are now also mandated to inform users every three months about the rules and the potential penalties for misuse of AI, which can include account suspension or termination.
