Even as the compliance regime for new takedown rules on unlawful content nears implementation, Big Tech major Meta on Tuesday said the proposed requirements—such as a three-hour window for acting on government takedown requests—could be difficult to operationalise. The company, however, reiterated its commitment to complying with the laws of the country.
“Operationally, three hours is going to be really challenging,” Rob Sherman, Deputy Chief Privacy Officer and head of AI policy at Meta, said during a media interaction at the India AI Impact Summit.
Sherman explained that each request requires due diligence before action. “Whenever we get a request from the government, we will have to look into it, we’ll have to investigate it, and validate it ourselves. So that’s just something that takes some amount of time,” he said. Sherman added that “this is an example where I think we’re concerned that had they come to us and talked to us about it, we would have talked about some of the operational challenges.”
Compliance Crunch
The comments come amid broader regulatory developments, including the rollout of the final rules for the Digital Personal Data Protection (DPDP) Act and ongoing discussions around intermediary responsibilities and platform accountability.
On India’s data protection framework, Sherman said Meta is still evaluating compliance requirements, particularly given the tighter timelines in India compared with norms in other regions. He added that existing compliance with European regulations does not eliminate the need for adjustments.
Nevertheless, Sherman said India remains a key market for Meta’s artificial intelligence (AI) initiatives, which form an integral part of its global operations.
India as an AI Powerhouse
“India is currently our number one market for Meta AI,” Sherman said, noting that the country has also been an early adopter of the company’s open-source Llama models and related tools.
Looking ahead, Meta is focusing on building what it calls personal superintelligence—AI systems tailored to individual users.
To support this, the company is investing in more efficient AI models, including those capable of running on-device, as well as developing techniques to deliver personalised outputs without significantly increasing computing costs.