The Automated Loophole Emptying Out Shreya Singhal

The Four Words That Made Automated Censorship Possible Section

79(3)(b) of the IT Act strips a platform of safe harbour immunity if it fails to act "expeditiously" upon receiving actual knowledge that a court or government has found content unlawful. In Shreya Singhal v. Union of India (2015), the Supreme Court read this provision narrowly precisely to prevent platforms from becoming private censors on vague suspicion: "actual knowledge" meant a specific order, properly issued, that someone at the company had actually read and assessed.

That single phrase - actual knowledge - is the entire legal foundation of today's controversy, because Meta's newly integrated API with the Home Ministry's Sahyog portal removes the "someone" from that sentence altogether.

What "Actual Knowledge" Was Built to Require Shreya

Singhal did not just narrow when a takedown obligation kicks in - it built a human checkpoint into the law's architecture. A platform was expected to receive an order, evaluate whether it came from an officer of the prescribed rank, check whether it stated adequate reasons and then act.

That evaluation step is precisely what gave "actual knowledge" its meaning as a conditional duty, not an automatic one. Apar Gupta of the Internet Freedom Foundation put this precisely: "an API that executes takedowns automatically has no knowledge at all."

Why the 2-3 Hour Deadline Made the Loophole Necessary The February 2026

amendment compressing the compliance window to 2-3 hours did not, on paper, touch Section 79(3)(b) or its "actual knowledge" standard. But it made manual compliance functionally impossible at the current order volume - nearly 1,275 blocking orders a day cannot be individually read, assessed for the issuing officer's rank and actioned within hours by a human review team.

Automation was not a deliberate evasion of Shreya Singhal so much as the only way to survive a deadline that was set without regard to the safeguard it was quietly dissolving. That is the real story: the erosion of judicial due-process protection here did not require a new law overturning the 2015 verdict - it only required a compliance timeline tight enough to make the human step commercially unaffordable.

What This Changes Going Forward

Once takedown becomes machine-to-machine, the burden of proof effectively inverts. Instead of the state having to justify a blocking order to a human reviewer who might push back, the platform's only remaining check is post-hoc - a user contesting a removal after the fact, if they even learn who ordered it, since Meta does not disclose the requesting authority in India.

For a UPSC aspirant, the exam-relevant insight isn't that blocking orders increased - it's that a procedural safeguard can be functionally erased without a single word of the underlying statute changing, simply by making the safeguard's cost exceed what a private company will pay to preserve it.