On 4 November I wrote that accountability for what gets amplified must flow upward, to the platform, not down to the teenager who posted. Two days later MeitY is reportedly weighing exactly that: safe harbour for carriers only, algorithmic liability for the rest, plus a direct user route to remove deepfakes and intimate images. I endorse the principle and flag the capture risk already visible in October's protest blocking, then propose the design — reach-based duties, a two-hour NCII clock, 48 hours for deepfakes, published amplification logs, a statutory appeal, and no blanket publisher status.
Two days ago I ended a post on smart glasses — the kind that turn any teenager into a broadcast station — with a line I have used before: accountability flows upward, not downward. The platform that decides a deepfake reaches a million people should answer for that reach, not the teenager who uploaded it. I did not expect MeitY to move this fast.
The Indian Express reported on 5 November 2026 that the IT Ministry has begun discussions on withdrawing Section 79 protection from platforms that use algorithms to rank, recommend or amplify user content, while keeping it for services that merely carry information: ISPs, telecom operators, peer-to-peer messaging. "They definitely determine what gets shown to users, and are no longer simple carriers of content," a senior official told the paper. Medianama reported the same day a second proposal: let a user approach a platform directly to remove a deepfake or a non-consensual image, without a police complaint or a court order.
I have argued the principle before. In April 2026 I wrote about draft amendments extending publisher-style remedies to user-posted news, and warned that vague functional tests are exactly where discretion enters. The instinct here is better; the drafting risk is the same.
The caution is real. The Indian Express itself records that Instagram's role in July's students' protests played a part in the ministry's thinking. Medianama notes what followed MeitY's 8 October advisory asking platforms to prioritise government complaints and label "misleading content": Meta restricted journalists' accounts and protest posts, and in November the Software Freedom Law Centre and creators alleged that protest content was being suppressed through keywords. If immunity turns on a ministry's satisfaction with your moderation, quiet suppression becomes the cheapest option — and a shadow-ban leaves no order to appeal.
The Supreme Court saw it on 15 October in Just Rights for Children Alliance v S Harish: loss of safe harbour, it said, is too drastic for routine non-compliance and too weak a deterrent. It asked for proportionate penalties, with that loss kept as a last resort.
So let any amendment follow the Court's logic. What I would suggest:
- Liability that attaches to reach, not posting: duties begin when a platform's ranker carries a post past a defined audience threshold. A video nobody sees harms nobody.
- Two clocks, not one: two hours for non-consensual intimate imagery, 48 hours for a reported deepfake, with interim labelling while the claim is checked.
- Published amplification logs: for any post an algorithm pushed, show what was promoted, when, and at whose request — including the government's.
- A statutory appeal to a tribunal, with a stay, not only an in-house grievance officer.
- No blanket publisher status: fines first, safe-harbour loss only for serious or repeat failure.
None of this is beyond MeitY, though amending the IT Act means Parliament and time. But after years of advisories, a statute that puts the duty where the reach is would be a real step. My only question is who decides when accountability flows upward — and where it is allowed to stop.
With regards,
Hemen Parekh
www.Iam-Immortal.ai / www.IndiaAGI.ai / www.YourContentCreator.in
6 November 2026
No comments:
Post a Comment