Intermediaries given new due diligence obligations as India proposes rules on synthetic media 

On 22 October 2025, the Ministry of Electronics and Information Technology (MEIT) released the draft Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Amendment Rules 2025 for public consultation, proposing updates to the existing 2021 Rules. The proposed amendments mark a significant step forward in India’s digital regulatory landscape by explicitly addressing the use and misuse of AI-generated or synthetically produced content.

By way of background, the existing rules prescribe due diligence obligations that intermediaries must implement in order to claim safe harbour protection under Section 79 of the IT Act 2000 for third-party content hosted on their platforms.

Definition of synthetic information

The draft rules define ‘synthetically generated information’ as information that is “artificially or algorithmically created, generated, modified or altered using a computer resource, in a manner that such information reasonably appears to be authentic or true”. This definition encompasses a broad range of AI-generated, manipulated or deepfake content, expanding the regulatory net to include misinformation, impersonation and algorithmically produced material that could mislead users.

Importantly, any reference to “information” under the rules – particularly where it concerns unlawful acts – will now be construed to include synthetically generated information, ensuring that AI-generated content is subject to the same due diligence and liability framework as other forms of digital information.

Applicability

The draft rules apply to intermediaries, including significant social media intermediaries (SSMIs).

Under the IT Act, an intermediary is any entity that, on behalf of a third party, receives, stores or transmits data or provides services related to such data. This broad definition includes web-hosting services, search engines, online marketplaces, telecom service providers and social media platforms.

An ‘SSMI’ is defined as a social media intermediary with more than five million registered users in India, which may include platforms such as Meta, YouTube and X.

Key obligations for intermediaries

The draft rules introduce detailed due diligence obligations for intermediaries that offer computer resources enabling the creation or modification of synthetic media. Such intermediaries must ensure that all synthetically generated content is prominently labelled or embedded with a permanent unique metadata or identifier that:

• enables immediate identification of synthetic media;
• is visibly displayed or made audible in a prominent manner – covering at least 10% of the visual surface area or, for audio content, the first 10% of its duration; and
• cannot be modified, suppressed or removed by users.

This requirement is designed to make AI-generated material instantly recognisable to audiences and regulators, addressing concerns around the proliferation of deepfakes and deceptive synthetic content.

Additional obligations for SSMIs

In addition to this, SSMIs must adopt stronger verification and disclosure mechanisms. Specifically, they must:

• require users to declare whether uploaded information constitutes synthetic media;
• deploy technical measures, including automated tools, to verify the accuracy of such declarations; and
• clearly display a label or notice indicating that the content is synthetically generated, where applicable.

If an SSMI knowingly permits, promotes or fails to act upon unlabelled synthetic media in violation of these provisions, it will be deemed to have failed in its due diligence obligations, potentially jeopardising its safe harbour protection under the IT Act.

Safe harbour and due diligence

To encourage the prompt removal of harmful AI-generated content and allay fears of liability connected with wrongful takedowns, the draft rules clarify that intermediaries removing or disabling access to synthetically generated information as part of reasonable efforts or upon receipt of valid grievances will not lose their safe harbour immunity. 

Policy context and next steps 

The draft rules emerge against the backdrop of a rapid rise in cases involving synthetically generated media used for impersonation, fraud and reputational harm – from celebrity deepfakes to market manipulation through fabricated statements. While courts have been developing jurisprudence around such misuse, the draft rules represent a structured regulatory response, aligned with international practice.

However, grey areas remain. The applicability of the draft rules to AI-based content-generation platforms (eg, chatbots or generative-AI applications) remains ambiguous, since such platforms may act as originators or deployers rather than intermediaries. Further, the definition of ‘synthetically generated information’ is broad, potentially capturing even benign uses such as photo filters or AI-assisted medical imaging. In their current form, the draft rules make no risk-based distinction, thereby subjecting creative and diagnostic tools to the same labelling obligations as deepfake-generating systems.

The MEIT invited public feedback until 6 November 2025, and stakeholders were expected to comment on the scope of intermediary liability for AI systems that generate content autonomously and the technical feasibility of embedding unalterable metadata identifiers. Once finalised, the draft rules will demand significant policy and technological adaptation. Their success will also hinge on a context-sensitive application.