Two days of news, one theme: the deepfake complaint has moved from op-ed to police station. Chiranjeevi and Ram Charan asked Amit Shah for strict action; in Ludhiana a burger vendor was booked for using AI images to extort money; VFX professionals want visible labels. India's February 2026 IT Rules already require labelling and metadata. What is absent is enforcement data: a 36-hour takedown deadline, a public compliance dashboard for every platform, and one national number that produces a pre-filled FIR.
At 3.37 this morning, The Tribune reported from Ludhiana that Model Town police have booked a burger vendor known as "Baba Burger Wala" on an FIR filed on 1 November. He is accused of creating AI-manipulated photos and videos of a gurdwara management official, posting them on Instagram, and then demanding money to remove them.
A day earlier, PTI reported (3 November, 00:30 IST) that Chiranjeevi and Ram Charan met Home Minister Amit Shah in Hyderabad. Per IANS (2 November), Shah advised a written representation, asked G. Kishan Reddy to coordinate, and assured urgent examination with stronger safeguards and strict penalties.
Same tool, two price points. One man blackmails a neighbour with a morphed face; the other is a megastar whose likeness is a market of its own. In Hindustan Times (3 November, 06:15 IST), VFX professional CB Arun Kumar says the flaw is not the label but its size — on Instagram the AI tag sits in small type under the username, "a tiny, hidden message". That is a design failure before it is a drafting failure.
The duties exist already. The IT Amendment Rules notified on 10 February 2026 brought synthetic content inside the intermediary framework, with AI-specific due diligence (Halverton & Co., 2 November 2026). TRAI has taken the sector-by-sector route: draft AI governance guidelines for telecom, published 30 October with comments due 27 November, requiring incident reporting to TRAI and accountability for AI supplied by third parties.
What is missing is a clock and a receipt.
- A 36-hour takedown deadline for synthetic impersonation, plus a public compliance page for each platform: labelling rate, median takedown latency, appeal outcomes — updated monthly, in the open.
- One national reporting number. An impersonation complaint should produce a pre-filled FIR draft with the platform's own logs attached, so no citizen needs a lawyer to report a fake.
- Liability that reaches the tool. When a generator turns a photograph into a blackmail threat, the maker of the generator answers alongside the uploader.
In March, in Grace for AI Labelling, I sketched the phased version: 30–60 days for public disclosure of method, 60–120 days of live pilots with third-party audits and public dashboards of labelling accuracy and takedown latency, then enforcement with agreed SLA windows. I wrote it for platforms buying time. It fits a burger vendor on Tajpur Road just as well.
In 2023, in Intentions are Good > Implementation has to be Great, I quoted the then-minister promising to make "the complaint and the FIR process completely digital" and to be "a party to the complaint". Three years later, the Ludhiana FIR was typed up the old way.
So here is the test I would apply to the memorandum going to Kishan Reddy. A celebrity complaint will get a fast lane; it always does. The design question is whether the same lane opens for a schoolteacher in Ludhiana with no lawyer and no publicity. Sunday's FIR shows the extortion market already exists at the bottom of the market. I believe 36 hours and one dashboard are affordable now. The alternative is another assurance.
With regards,
Hemen Parekh
www.Iam-Immortal.ai / www.IndiaAGI.ai / www.YourContentCreator.in
3 Nov 2026
No comments:
Post a Comment