The Ministry of Electronics and Information Technology (MeitY), on 15 March 2024, issued an advisory to all intermediaries and platforms regarding the deployment and hosting of artificial intelligence models, particularly generative AI tools. The advisory, issued under the framework of the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, imposes due diligence obligations on platforms to ensure that AI-generated or synthetic content is appropriately labeled before being made available to Indian users.
Background
The advisory emerged against the backdrop of increasing deployment of large language models and generative AI tools across Indian digital platforms. In the months preceding the advisory, concerns had mounted regarding AI-generated deepfakes, misinformation, and synthetic media being circulated without identification on social media and messaging platforms.
The IT (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, already imposed certain due diligence obligations on intermediaries under Rule 3. The advisory built upon this existing framework by extending the scope of intermediary responsibility to encompass AI-generated content specifically. Notably, the advisory was issued during a period of heightened political sensitivity, with the 2024 Lok Sabha general elections approaching, amplifying concerns about AI-enabled disinformation.
Key Provisions
The MeitY advisory set out the following requirements for intermediaries:
Mandatory labeling of AI-generated content: All intermediaries hosting or deploying AI models must ensure that content generated by AI systems -- including text, images, audio, and video -- is clearly labeled or watermarked as AI-generated or machine-generated before being made available to users.
Due diligence for AI tool deployment: Platforms making AI tools available to Indian users must exercise due diligence to ensure that such tools do not generate content that violates existing laws, including provisions relating to misinformation, defamation, and obscenity under the Information Technology Act, 2000.
Broad applicability: The language of the advisory was drafted broadly enough to encompass all AI tools, including generative AI chatbots, image generators, and text-to-video models, regardless of whether the platform developed the AI model itself or merely hosted a third-party tool.
Compliance timeline: Intermediaries were advised to implement labeling mechanisms and content moderation systems for AI-generated content, though the advisory did not specify binding penalties for non-compliance, relying instead on the existing enforcement framework under the IT Rules.
Government approval suggestion: An earlier version of the advisory had suggested that AI models require government approval before deployment, which was subsequently softened to an advisory framework following industry pushback.
Implications for Practitioners
This advisory represents a significant step in India's evolving approach to AI governance, operating through the existing intermediary liability framework rather than through standalone AI legislation. For technology companies, the immediate operational challenge lies in developing reliable technical mechanisms for detecting and labeling AI-generated content at scale -- a capability that remains technically imperfect across the industry.
Legal counsel advising technology platforms should note that while the advisory itself does not carry statutory penalties, non-compliance could jeopardise an intermediary's safe harbour protection under Section 79 of the IT Act, 2000, if the advisory is treated as part of the due diligence framework under the IT Rules. This indirect enforcement mechanism makes the advisory more consequential than its non-binding language might suggest.
The advisory also raises questions about the jurisdictional reach over AI models hosted outside India but accessible to Indian users, an area that remains unsettled and will likely require further regulatory clarification.