India adds deepfake labeling rules and tightens takedowns to 2–3 hour windows

India is moving from “platform policy” to formal regulation on deepfakes, forcing social networks to label synthetic media and respond to takedown demands in hours, not days—raising both compliance costs and content-risk for brands running influencer and UGC-heavy campaigns.

New compliance deadlines and deepfake provenance requirements

The government this week published amendments to India’s 2021 IT Rules that pull deepfakes into a defined regulatory framework and sharply compress response timelines. Platforms must comply with official takedown orders within three hours, and in some urgent user-complaint scenarios within two hours, according to the gazette notification (egazette.gov.in PDF)/ViewPDF.aspx)).

For platforms that host user-shared audio or video, the changes require users to disclose whether content is synthetic, and require services to deploy technical systems to verify those disclosures. Deepfakes must be clearly labeled and include traceable provenance data—metadata intended to make origin and edits auditable across reposts.

Several synthetic-content categories are outright prohibited, including deceptive impersonation and non-consensual intimate imagery. The bigger operational point for global platforms is liability: missing these deadlines can jeopardize “safe harbor” protections, increasing exposure for user-posted content.

What it means for marketers, creators, and moderation stacks

The rules implicitly push heavier automation in moderation and verification—more classifiers, watermark/provenance checks, and faster triage. That will affect AI video workflows, especially campaigns that remix creator footage, use synthetic voiceovers, or localize ads at scale.

Legal and civil-society observers are already warning that the timelines leave limited room for human review and may incentivize over-removal. Internet Freedom Foundation argued the “impossibly short timelines” reduce meaningful due process and could accelerate censorship (x.com statement).

The amendments take effect February 20, aligning with India’s AI Impact Summit dates in New Delhi (impact.indiaai.gov.in). For marketing and e-commerce teams, the near-term play is operational: tighten asset provenance, keep model-release documentation, and ensure agencies can rapidly provide source files when content is flagged.