Proposed DPDP timelines a challenge: Meta's Chief of Global Policy

The government’s proposal to trim the compliance deadlines for some of the provisions of the Digital Personal Data Protection (DPDP) Act, from the current 18 months to a year, poses practical challenges, such as shipping out updates to comply with the norms, Meta’s Chief of Global Policy Kevin Martin told Aashish Aryan in an in-person interview. Martin, who was in India to attend the AI Impact Summit, also said the AI talent base in India was unparalleled. Edited Excerpts:

 

What are your concerns about the proposal to shorten DPDP compliance timelines?

 

We do have some concerns with the timeline being shortened from 18 to 12 months. We think it would make more sense to leave it at the original timeframe… (18) months is already a short time frame in our view to adjust all of our systems and ensure we are in compliance. Ensuring compliance with the country’s rules, especially in a critically important market such as India, is a priority for us. But shortening the timeline is not as constructive.

 

The timelines for compliance with the requirement to take down synthetically generated information have also been shortened. What are your thoughts on that?

 

That is a much shorter timeline and almost immediate. Technically, that is very complex, and that is a particular challenge for us. We share the goal of addressing concerns about synthetically generated information.

 

We have systems in place that take down content when someone complains or the government sends us takedown requests. Some of the new requirements, however, require us to do something even before people can upload such content. That kind of system is much more burdensome and could be even more burdensome for consumers and users. We are talking to the government about it and sharing our concerns so we can come into compliance as quickly as possible.

 

What is Meta’s broader AI vision, especially in countries such as India?

 

There are several elements critical to our AI development, such as the right kind of talent, sufficient energy to power our computing, and access to data. Those are the key ingredients from our perspective. We are very encouraged and optimistic about India’s general approach to AI. The talent base is unparalleled in the country, and about 90 per cent of the startups had an AI component in what they were doing.

 

The talent ecosystem has been known here for a long time, but the focus on AI is significant. We think we need a regulatory environment that encourages innovation.

 

What do you think about the race to develop sovereign AI capabilities?

 

It is important for us to ensure that the AI services we provide are culturally relevant and accessible to the majority of people. That is one of the benefits of an open-source alternative approach. The sovereign model approach ensures that users have access to AI in a way they understand and that is relevant to them.

 

We think that it can be done without having a government or sovereign model per se. The ecosystem should be built in a way that facilitates startups building on top of it.

 

There are reports that Meta will also adopt a proprietary AI model approach, despite consistently suggesting it will opt for open AI models.

 

We have always viewed that there will probably be both proprietary and open AI models. Both are important to the ecosystem. There are certain things we will want to offer on a proprietary basis, but we are not trying to sell those models.

 

We have partnerships here in India as well, and some of those may develop into proprietary models over time. We are, however, going to continue to support the open source component as well. It will be a dual system.