Takedown requests on X against Indian accounts that document online censorship

On 31 October and 2 November, two notices bearing the name of the Delhi Police reached X, asking it to remove or withhold (hide content only in specific countries) three accounts that document online censorship. SFLC.in, an Indian digital rights organisation named in the notices, published the copies X forwarded to it and described the notices as "purportedly issued". As of 8 November, neither the Delhi Police nor the Indian government had publicly confirmed them. According to X's help page, withholding is limited to the country that made the request, so what readers outside India see is not affected.

It began on 29 October, when two Indian digital rights organisations and an X account that compiles blocking orders received a legal notice claiming to act for the RSS, a Hindu nationalist organisation, demanding that they stop documenting restricted content. According to Newslaundry, relaying IFF (the Internet Freedom Foundation, an Indian digital rights organisation), several RSS functionaries later questioned on X whether the notice was genuine. The two organisations posted the notice publicly on X on 30 October, and those two posts are what the first police notice targeted, on the grounds that they defamed the RSS.

According to SFLC.in's statement of 5 November, the court order fields in both notices were left blank, and the first cited section 356 of the Bharatiya Nyaya Sanhita (BNS, India's penal code), which covers defamation. The second notice listed the three accounts in their entirety, cited section 79(3)(b) of the Information Technology Act, 2000 and the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, and stated that lawful notices require action within three hours. Section 79 is the safe harbour provision (the protection that keeps platforms from being liable for what users post), and section 79(3)(b) means a platform that fails to remove content after receiving notice may lose that protection.

X forwarded copies of both notices to the affected organisations on 3 November, and according to IFF the emails did not say whether X had acted on them. Reuters reported on 4 November that both organisations' accounts were still viewable that afternoon, and that X, the Delhi Police, India's Ministry of Electronics and Information Technology (MeitY) and the RSS had not responded to its requests for comment. As of 8 November we could find nothing on later developments or responses, nor whether the notices had been submitted to a database of takedown notices such as Lumen.

SFLC.in checked where the notices came from. The IP address of the sending server is registered to the Delhi Police and is listed in the domain's SPF record (a public setting that lists which servers may send mail for a domain), yet emails sent to the two police addresses given in the notices bounced. Since the two findings point in different directions, SFLC.in did not draw a conclusion about authenticity from them. It asked the Delhi Police and X for clarification and, as of 5 November, had received no official reply.

SFLC.in also tested X's Law Enforcement Request form. X's Legal Request Portal sends a verification link to an official email address, but this form did not. On 4 November, using a made-up email address, SFLC.in submitted a request targeting its own post, and X logged it all the same. By that afternoon the form had been changed to send a verification code to the submitted address, and as of 8 November X had not publicly explained the flaw or the fix.

Perspective

Whether a notice is valid depends on the law it cites and on who submitted it, and the people affected can only appeal once they have been notified. Besides notices, India has another route under section 69A of the Information Technology Act, 2000, through which the central government blocks content by an order that records its reasons in writing. According to the Apple notice the developer made public, the removal of Bitchat from the App Store in India on 3 October cited section 69A (see Dispute over the removal of the Bluetooth mesh messaging app Bitchat in India).

Our inference is that a notice combined with a three-hour deadline leaves a platform very little time to judge whether a request is valid. Whether a form verifies the identity of the person submitting it therefore affects whether impersonated requests can enter a platform's handling process. SFLC.in asked X through which channel the two notices had been submitted and, as of 5 November, had no answer.

According to X's help page, X determines a user's country from location information it receives, and withheld posts and accounts carry a notice. One kind of notice means X withheld the content in response to a valid legal demand, such as a court order, and the other means X withheld it under local law in response to a report filed through specific support channels. Both notices in this case are titled Illegal Content Report, so our inference is that if the content were withheld, it would more likely carry the local-law notice. X has not said how it classified these requests.

The same help page says that, unless prevented by something like a sealed court order, X notifies affected people by email to the address registered on the account or in the app. The notification usually includes a copy of the request, and people can appeal by replying to the email where content was withheld on a legal demand, or through the Help Center where it was withheld under local law. We infer from this that an organisation whose registered inbox goes unread may not learn that its account has been withheld, and may not receive the copy of the request. People who run organisation accounts in India can check the registered inbox regularly and, if X notifies them, appeal through the channel given in the notification.

On the defamation offence the notices cite, SFLC.in's analysis is that criminal defamation under the BNS is a non-cognisable offence, so police need a magistrate's order before acting. SFLC.in also argues that the officials the Delhi government authorised in June to issue such notices do not include the post that signed them, and that restricting entire accounts rather than specific content fails the tests of necessity and proportionality (this is the organisation's analysis, not a finding by a court or the government). Reuters added the Indian government's general position, not specific to this case, that its directives to platforms target unlawful or misleading content and respect lawful expression.

According to Inc42 and SFLC.in, relaying a report by the Indian Express, MeitY has begun discussing whether to reconsider safe harbour for social media platforms. As relayed by Inc42, the report says the government is studying whether to limit the protection to internet service providers, telecom operators and peer-to-peer messaging platforms, and that the discussions are still exploratory. SFLC.in relays the reasoning that platforms curate and amplify content with algorithms and so are already publishers. MeitY had made no public statement as of 8 November, and linking the police notices to the safe harbour discussion is our own connection, because both involve section 79.

Citing Supreme Court precedent, SFLC.in argues that section 79 is a statutory protection, not a privilege that executive discretion can withdraw. It also warns that if platforms become liable for user content, political speech, journalism and criticism of officials are the most likely to be over-removed. In September, MediaNama analysed what government sources had said about Meta and concluded that the position then rested entirely on anonymous briefings, and that safe harbour is a defence platforms raise when sued, for a court to decide. That analysis predates the November Indian Express report, but the sources' statements it examined are similar to that report, and placing the two side by side is our own connection.

We compared how takedown requests reach platforms in three other places in Asia. In Hong Kong, Schedule 4 to the Implementation Rules for Article 43 of the National Security Law lets a designated officer, authorised by the Commissioner of Police in national security cases, require a platform to take a disabling action on a message without a magistrate's warrant, and the schedule does not say the publisher must be given reasons or a route of appeal. In Taiwan, Article 46 of the Protection of Children and Youths Welfare and Rights Act requires a platform told by the competent authority that content harms children to restrict children's access or remove the content, with no court procedure, though the provision protects children only. In mainland China, Article 52 of the Cybersecurity Law, as amended in 2025 and in force since 1 January 2026, has the state internet regulator and other authorities order network operators to stop transmission and have content from outside China blocked, and we found no requirement in the laws we read to go to a court first.

Transparency reports are one place where such requests appear. According to Hong Kong Free Press in 2023, citing Google's transparency report, the Hong Kong government asked Google to remove 330 items in 2022, 57 of them related to national security, up from 116 items in 2021. These are 2022 figures, not current ones.

Readers outside India do not need to change any settings. Points to watch next are whether X withholds these accounts in India, whether the Delhi Police or MeitY responds publicly, and whether the safe harbour discussion becomes a formal amendment or draft rules.

In its article of 6 November, EFF called on platforms to be transparent about the requests they receive, to notify affected users, and to provide avenues for appeal against restrictions. When comparing platforms, readers can also look at whether a platform verifies the identity of whoever submits a request and whether notices come with a court order. Since the government, as relayed by Reuters, says its directives target unlawful and misleading content, one question is whether platforms can tell, within the deadline, whether a notice is genuine and lawful. We found no public assessment of whether, once safe harbour is narrowed, platforms would remove unlawful content more precisely or remove more lawful speech.