India should reconsider its proposed regulation of online content

The article was published in the Hindustan Times on January 24, 2019. The author would like to thank Akriti Bopanna and Aayush Rathi for their feedback.

Flowing from the Information Technology (IT) Act, India’s current intermediary liability regime roughly adheres to the “safe harbour” principle, i.e. intermediaries (online platforms and service providers) are not liable for the content they host or transmit if they act as mere conduits in the network, don’t abet illegal activity, and comply with requests from authorised government bodies and the judiciary. This paradigm allows intermediaries that primarily transmit user-generated content to provide their services without constant paranoia, and can be partly credited for the proliferation of online content. The law and IT minister shared the intent to change the rules this July when discussing concerns of online platforms being used “to spread incorrect facts projected as news and designed to instigate people to commit crime”.

On December 24, the government published and invited comments to the draft intermediary liability rules. The draft rules significantly expand “due diligence” intermediaries must observe to qualify as safe harbours: they mandate enabling “tracing” of the originator of information, taking down content in response to government and court orders within 24 hours, and responding to information requests and assisting investigations within 72 hours. Most problematically, the draft rules go much further than the stated intentions: draft Rule 3(9) mandates intermediaries to deploy automated tools for “proactively identifying and removing [...] unlawful information or content”.

The first glaring problem is that “unlawful information or content” is not defined. A conservative reading of the draft rules will presume that the phrase means restrictions on free speech permissible under Article 19(2) of the Constitution, including that relate to national integrity, “defamation” and “incitement to an offence”.

Ambiguity aside, is mandating intermediaries to monitor for “unlawful content” a valid requirement under “due diligence”? To qualify as a safe harbour, if an intermediary must monitor for all unlawful content, then is it substantively different from an intermediary that has active control over its content and not a safe harbour? Clearly, the requirement of monitoring for all “unlawful content” is so onerous that it is contrary to the philosophy of safe harbours envisioned by the law.

By mandating automated detection and removal of unlawful content, the proposed rules shift the burden of appraising legality of content from the state to private entities. The rule may run afoul of the Supreme Court’s reasoning in Shreya Singhal v Union of India wherein it read down a similar provision because, among other reasons, it required an intermediary to “apply [...] its own mind to whether information should or should not be blocked”. “Actual knowledge” of illegal content, since then, has held to accrue to the intermediary only when it receives a court or government order.

Given the inconsistencies with legal precedence, the rules may not stand judicial scrutiny if notified in their current form.

The lack of technical considerations in the proposal is also apparent since implementing the proposal is infeasible for certain intermediaries. End-to-end encrypted messaging services cannot “identify” unlawful content since they cannot decrypt it. Internet service providers also qualify as safe harbours: how will they identify unlawful content when it passes encrypted through their network? Presumably, the government’s intention is not to disallow end-to-end encryption so that intermediaries can monitor content.

Intermediaries that can implement the rules, like social media platforms, will leave the task to algorithms that perform even specific tasks poorly. Just recently, Tumblr flagged its own examples of permitted nudity as pornography, and Youtube slapped a video of randomly-generated white noise with five copyright-infringement notices. Identifying more contextual expression, such as defamation or incitement to offences, is a much more complex problem. In the lack of accurate judgement, platforms will be happy to avoid liability by taking content down without verifying whether it violated law. Rule 3(9) also makes no distinction between large and small intermediaries, and has no requirement for an appeal system available to users whose content is taken down. Thus, the proposed rules set up an incentive structure entirely deleterious to the exercise of the right to freedom of expression. Given the wide amplitude and ambiguity of India’s restrictions on free speech, online platforms will end up removing swathes of content to avoid liability if the draft rules are notified.

The use of draconian laws to quell dissent plays a recurring role in the history of the Indian state. The draft rules follow India’s proclivity to join the ignominious company of authoritarian nations when it comes to disrespecting protections for freedom of expression. To add insult to injury, the draft rules are abstruse, ignore legal precedence, and betray a poor technological understanding. The government should reconsider the proposed regulation and the stance which inspired it, both of which are unsuited for a democratic republic.