An Indian Court Ordered a Political Deepfake Removed. It Had Already Done Its Work.
What happened
In December 2025, a fabricated video began circulating across social media in India, showing Prime Minister Narendra Modi and industrialist Gautam Adani in a conversation that never happened. The video, built using deepfake technology, depicted the two men discussing corruption in terms designed to damage their reputations and, the court later found, to mislead ordinary voters and international investors alike. The content was shared through political channels before any platform or regulator had flagged it as synthetic.
An Ahmedabad court ordered the video taken down and directed X (formerly Twitter) and Google to remove it within 72 hours if the parties responsible had not done so themselves. The court described the content as carrying malicious imputations capable of causing irreparable injury to the Adani Group's reputation. It also cited risks to public order and stakeholder trust. By the time the order was issued, the video had already achieved wide circulation. The legal mechanism worked as designed; the timing problem it exposed did not go away.
The incident is a product of what the record describes as the democratization of generative AI tools, which now allow convincing synthetic media to be produced in minutes at negligible cost. The technical barrier to creating a believable deepfake has effectively collapsed. What has not kept pace is the detection and disclosure infrastructure that would allow a viewer, a platform, or a court to quickly establish whether a video is synthetic before it spreads. The video circulated without any embedded provenance signal, no watermark, no metadata standard, nothing that would have marked its origin at the moment of publication.
Platform moderation added another layer of delay. Even after a formal court order, the enforcement window gave platforms 72 hours to comply, a span in which a viral video can reach millions of viewers and generate thousands of reshares across jurisdictions where the order has no force. The record explicitly notes that moderation lag times are insufficient for viral deepfakes. No mandatory disclosure requirement compelled whoever created the video to label it as synthetic, and no requirement compelled political actors circulating it to identify where it had come from.
What this case illustrates, beyond any single court ruling, is the absence of a baseline verification layer at the point of publication. The video's synthetic origin became actionable only after the fact, through litigation, not at the moment it was first shared. A provable record of what a system produced, when, and under whose direction would have made the fabrication visible the moment it was distributed rather than weeks into its circulation. Without that record, deepfake political content can complete its reputational work before any institution possesses the information needed to respond.
Reported impact
- Affected parties
- Not publicly disclosed
- Harm type
- Not publicly disclosed
- Scale
- Not publicly disclosed
- Financial impact
- Not publicly disclosed
- Regulatory action
- Not publicly disclosed
Classification
Relevant governance controls
Governance control mapping is not available for this record.
- No controls mapped
Not publicly disclosed
Control mapping is analytical. It does not state that any control would have prevented the incident.
Sources and evidence
This record was researched and written by the Index. The event is also catalogued in the following database, which is listed for cross-reference.