Automated content moderation by platforms has become a tool of de facto censorship. Suggest a rights-respecting framework for content takedowns in India. (10M)

In October 2026, posts and whole accounts on Instagram, Facebook and X began showing "Your post is unavailable in India". The notices said the content was restricted by an automated system under the IT Rules, 2021 [1]. When an algorithm removes speech without giving a reason, the free speech right under Article 19(1)(a) is cut back without the checks that Article 19(2) demands.

Why automation turns into de facto censorship

• Shorter deadlines: In 2026, the time to remove content flagged by a government or court order was cut from 36 hours to 3 hours [2]. That leaves no time for a person to review it, so platforms leave the decision to algorithms.
• One-sided incentives: A platform that misses a deadline risks losing safe harbour under Section 79. Wrongly removing a lawful post costs it nothing. So platforms remove whenever they are unsure.
• Pressure from advisories: During the anti-SIR protests, MeitY's 8 October advisory asked platforms for "enhanced diligence" on ongoing public events [1]. Advisories are formally non-binding, but platforms treated it as an order.
• Secrecy: Blocking keeps growing. MeitY directed the blocking of 9,845 URLs in 2024 up to October [3]. Yet the notices do not say who ordered a removal or why [1].

A rights-respecting framework

1. Legality: Route every government order through Section 69A. In Shreya Singhal (2015), the Supreme Court upheld that section because it has committee review and written reasons. If advisories are to become binding, Parliament should debate it, not the executive through rules.
2. Different deadlines for different harms: Keep the 3-hour window for deepfakes, impersonation and incitement to violence, which the 2026 amendment was mainly aimed at [2]. Give political and news content a longer window with human review.
3. Notice with reasons: Every restriction should say whether it came from a government order, a court order or platform policy, with an order number. UNESCO's 2023 Guidelines ask for exactly this [4].
4. Accountability for algorithms: Independent external audits of moderation tools, plus India-specific reports on how many posts were removed and later restored [4].
5. Effective remedy: Time-bound appeals to Grievance Appellate Committees, human review before any account is suspended, and disclosure of the blocking order to the affected user.

Fast action against deepfakes and transparency for citizens can go together; the real problem is speed without a record. A framework built on legality, proportionality and due process, in line with Shreya Singhal and UNESCO's guidelines, would let India curb real online harm. It would also keep Article 19(1)(a) a working right, especially during elections.