MeitY’s advisory on deployment of AI models

Please click here to download the Prism as PDF.

On March 1, 2024, the Ministry of Electronics and Information Technology (“MeitY”) issued an advisory (“Advisory’) directing all intermediaries and platforms to label any under-trial/unreliable artificial intelligence (“AI”) models, and to secure explicit prior approval from the government before deploying such models in India. This Advisory follows a strong response by MeitY considering the recent Google-Gemini row and also builds on an earlier advisory dated December 23, 2023 (“December Advisory”) specifically targeting the growing concerns around propagated by AI- Deepfakes and mandating communication of prohibited content to the users.

Provisions of the Advisory

The Advisory states that all intermediaries and platforms are to ensure that their AI-models, large language models(LLMs), generative AI/software(s) and algorithm(s) or computer resource does not permit any discrimination or threaten the integrity of the electoral process and to prohibit their users from contravening the provisions of the IT Act 2000 and the IT (Intermediary Guidelines and Digital Media Ethics Code) Rules 2021 (“Rules”) which does not permit hosting, displaying, modifying, publishing transmitting, storing, updating or sharing unlawful content.

Further, the Advisory also states that the use of under-testing/unreliable AI models of intermediaries and platforms on Indian internet is subject to prior explicit approval from the government and its deployment is dependent on a due diligence of the AI model’s possible and inherent unreliability of the output generated which is to be informed to its users by way of a consent mechanism essentially containing the risks and consequences of dealing with such unlawful information. The Advisory has also requested AI platforms and intermediaries using its software or any other computer resource in such a manner to generate information that could be misused or considered deepfake to label such information with a unique metadata or identifier in a manner that such label, metadata or identifier can be used to identify that such information has been generated by the AI system of the intermediary, identify the intermediary and the creator or first originator of such misinformation or deepfake.

Moreover, this Advisory, very similar to the December Advisory, highlights the possibility of severe penal consequences to intermediaries, platforms and users in the event of non-compliance of the IT Act, Rules and criminal laws.

Clarifications on the Advisory

Further, the Minister for Electronics and Information Technology, Mr. Ashwini Vaishnaw and Minister of State, Mr Rajeev Chandrasekhar have confirmed that this Advisory is not binding and only encourages voluntary compliance to prevent legal action by consumers.

Mr Rajeev Chandrasekharan has clarified that the Advisory is intended for significant /large platforms, not AI start-ups are required to seek prior approval from the government.

Conclusion

Although under Section 13 of the Rules, MeitY can issue appropriate guidance and advisory to publishers, it is unclear if MeitY is within its scope of the Rules to issue advisories specific to AI governance, thereby questioning its validity. The advisory, by its very nature, is not binding as held by a plethora of judgments of Indian Courts.

The threshold for determining “significant/large platforms” and “start-ups” remains unclear. The parameters for evaluating “under-tested” and “unreliable” AI are not defined, thereby making voluntary compliance difficult.

This Prism has been prepared by: