Exclusive: AI Models May Need To Pass Tests Before Launch Under India's Draft Rules

The consultation paper may provide a 60-day feedback window, while sectoral regulators such as RBI, SEBI and TRAI could introduce tailored compliance requirements.

Advertisement
Read Time: 2 mins

India is moving closer to putting a regulatory framework in place for artificial intelligence, with the Ministry of Electronics and Information Technology (MeitY) expected to release a consultation paper on the proposed AI rules as early as next month, NDTV Profit has learnt  exclusively from sources familiar with the matter.

The draft AI law is in its final stages, with the government weighing a separate set of safety requirements for the most powerful AI models.

Advertisement

The consultation paper is expected to seek feedback from industry stakeholders and the public before the government moves on to formal legislation.

MeitY may give a 60-day window for public consultation, sources told NDTV Profit. Feedback received during this period will likely play a key role in determining the final shape of the proposed regulations and the timeline for their implementation.

Advertisement

The government is also considering issuing a formal legislation in the Budget Session. The timing and the scope of the final framework will depend on the responses to the consultation process.

Developers, platforms and deployers of AI could face greater regulatory oversight under the rules being proposed, especially in cases where the artificial intelligence systems cause harm.

Advertisement

Concerns over deepfakes, AI-enabled fraud and impersonation have emerged as major challenges with the rapid adoption of generative AI tools and the framework is expected to address them.


Separate Safety Standards for Advanced AI Models

One of the key areas under consideration is the possibility of introducing separate safety requirements for frontier AI models, which are among the most advanced and powerful artificial intelligence systems.

Sources also noted that the government is looking at defining frontier AI models based on factors such as computing power and capabilities.

This could help distinguish highly advanced systems from smaller AI applications and ensure that regulatory requirements are proportionate to the risks involved.

Advertisement

The proposed framework may also require AI developers to conduct mandatory safety testing before releasing their models to the public.

Sandbox trials could be another part of the proposed approach, allowing developers to test AI systems in controlled environments before wider deployment.

Such requirements could be particularly relevant for models capable of generating realistic content, operating autonomously or carrying out complex tasks with limited human intervention.

(This is a developing story)

Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.


Loading...