India Mandates AI Content Labelling, Platform Accountability

India's Ministry of Electronics and Information Technology (MeitY) issued a comprehensive advisory on 1 March 2024. This directive targets the use and dissemination of artificial intelligence (AI) models. It applies to intermediaries and platforms operating within India.

The advisory specifies that AI models, software, or algorithms must not generate content that is unlawful. This includes content prohibited under the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021. Such content must be identifiable as a potential deepfake or misinformation.

Platforms are now required to ensure their AI models do not produce or permit users to create specific categories of prohibited content. These categories include content that is biased, discriminatory, or threatens electoral integrity. Content undermining national security or public order is also prohibited.

Critically, the advisory mandates explicit permission from the government before deploying any 'under-testing/unreliable' AI models. This requirement applies to models within the 'significant social impact' or 'risk of harm' categories. The advisory does not define these categories. Platforms must also ensure these models do not contain inherent biases or discriminate. They must also undergo due diligence before deployment.

All AI-generated content must be accompanied by a clear and indelible identifier. This includes metadata or a unique mark. It must indicate the content was synthetically produced. This measure aims to prevent misidentification of AI-generated media as human-created.

The directive also requires platforms to inform users about the potential fallibility and unreliability of AI models. Users must acknowledge and accept these terms. This consent mechanism applies before any AI model is deployed or made accessible. The advisory became effective immediately upon its release. It expands upon prior advisories concerning deepfakes.

The Indian government's move signifies a proactive regulatory stance on AI governance. It prioritises user awareness and content authenticity. How will platforms adapt their deployment and content moderation strategies to meet these new requirements?