2-hour deepfake takedown timeline lands Big Tech in compliance furore

The Centre has maintained that most technology firms are well-equipped to automate scanning and curb harmful content online. However, lawyers and policy executives said sweeping automation could lead to legitimate posts being taken down, content creators and advertisers being disrupted, and users potentially taking hours to make a single post—each a factor that may break the smooth flow of operations on the open internet.

Such a short window for taking down content is rare. The US, which signed its ‘Take It Down Act’ into law on 19 May last year, mandates the removal of non-consensual sexual imagery within 48 hours of the matter being reported by a victim.

The European Union (EU) AI Act urges platforms to remove content proactively, but without a specific timeline. The same applies in China too, with its law mandating “high pressure” on social media firms to remove illicit AI-generated or modified deepfakes—but without specifying a timeline.

Big Tech firms that run Instagram, LinkedIn, X (formerly Twitter) and YouTube—India’s top social media platforms—are currently on a wait-and-watch mode with the new timelines for content takedown in India.

While the government removed a contentious proposal of watermarking 10% of practically any online content, it sharply tightened timelines for taking down objectionable material from digital platforms.

Non-consensual sexual imagery—including deepfakes—must be removed within two hours instead of 24 hours previously and any other unlawful content is to be removed within three hours instead of 36 hours.

Compliance burden

Rohit Kumar, co-founder of The Quantum Hub, a public policy advocacy firm that works with the top technology companies, said the new timelines will “make it difficult” for companies to achieve compliance.

“While removing the prescriptive proposition of a 10% watermark on AI-generated content is a progressive move, the timeline impositions could make it significantly difficult for intermediaries, both small and large, to match the expected burden of compliance,” Kumar said. “This is because while most content-monitoring algorithms are automated, taking down specific content based on reports will inevitably require human oversight. Compressing them exponentially will be a huge addition of cost to the large tech companies and could be nearly impossible for the smaller ones.”

The notified policy “is like taking one step forward, but two steps backward,” a senior executive at one of the top three Big Tech firms said, requesting anonymity. “Progressive economic destinations such as Japan and Singapore have previously deliberated on hard-touch regulation of AI but eventually backed down from it because AI is a nascent, evolving field—and a hard crackdown in terms of regulation could be difficult to implement.”

The significantly narrowed timelines could mean that India will now become one of the most heavily monitored and censored social media destinations globally, the executive said.

Rutuja Pol, a partner at law firm Ikigai, concurred.

“To start with, the 10-day window given to firms to comply with the new rules itself is difficult. While tech firms do have strong technical capabilities, legal frameworks in a country of India’s scale invariably require longer compliance windows for companies of all sizes to prepare for it,” Pol said.

She added that the shortened timelines cannot be matched entirely with automation.

“For small social media platforms, the cost of compliance may just be so high that it becomes infeasible for them to continue operating in India,” Pol said.

Queries sent to Meta, Google, YouTube and OpenAI did not receive responses until press time.

New Delhi, however, is confident of its law. A senior official with direct knowledge of the proceedings said the ministry of electronics and IT had several discussions with industry stakeholders to arrive at this law, and there were no indications that the tech firms did not have the capabilities to comply with what they have been asked to do.

The official added that while the ministry has heard “extensive and exhaustive feedback” on the rules, “we did not really need to as this is a balanced law that is not asking for anything unreasonable.”