It has been over ten years since China irrevocably changed the nature of the internet. It did this by coercing major American internet platforms to either enable Chinese government censorship and surveillance or leave China. There are two major differences between China and India in this context: India has a history of a stronger commitment to human rights, while China had a rich local ecosystem of its own web-based platforms. Together they mean that our experience of cutting ourselves off from the global Internet would be quite different from China’s. This is what is being contemplated by the Information Technology ministry’s proposal to amend the rules under the Information Technology Act.
If the draft amendment is passed, it will disconnect us from much of what the global internet has to offer, just as Chinese residents were disconnected. This includes the web-based applications on our phones. Unlike China, we do not have home-grown alternatives that compare in sophistication to the global platforms. This should offer reasons for concern even to those who do not value our fundamental rights. Those of us who are concerned about human rights should worry about this amendment, and about the background of subtle undermining of rights online within which it has emerged.
Amending the IT Rules
The proposal published by the Information Technology ministry on Christmas Eve will undermine freedom of expression and privacy online. This is a draft amendment of the Indian Information Technology rules that I refer to as the Purdah amendment since it would draw a veil over the Indian internet. If it becomes law, the Purdah amendment will result in unprecedented censorship.
To understand the enormity of its impact, it is necessary to analyse and read the amendment in detail, and consider it in combination with other Indian laws and law enforcement practices. This piece confines itself to two significant and worrying changes that the amendment contemplates. The first is the requirement of incorporation in India for certain online intermediaries. The second is the proactive, automated removal of unlawful speech by companies.
Incorporation in India: The most extreme feature of the proposal might be that it requires companies with more than 50 lakh users in India to incorporate in India. It also enables the Indian government to require other companies to incorporate in India. If global companies are unable to incorporate here, we potentially lose access to valuable content, such as information from Wikipedia.
This violates our right to receive information under Article 19(1)(a) of the Indian Constitution, but is only one part of the problem raised by forced incorporation. The other part is that if companies do incorporate in India, their physical presence in the country will make them vulnerable to unreasonable requests from the Indian government.
A company becomes more amenable to removing content or enabling government surveillance if its staff are intimidated with income tax investigations, arrests and the many other forms of pressure that may be exerted on people who live within the country. Platforms under such pressure have reduced incentive to refuse unreasonable government demands to censor or share users’ private information.
If web-based platforms start acting on government demands that violate out rights to privacy and freedom of expression, we will have little scope of appeal to the judiciary since we are unlikely to be notified of the specific government order.
Automated pre-censorship: The second troubling feature of the amendment, is that it will make social media companies use algorithms to “proactively” identify and disable public access to “unlawful information or content”. This means that companies will deploy heavy-handed automated pre-censorship tools for everything ranging from defamation to child pornography, resulting in extensive pre-censorship all round.
When Indian cable TV networks were required to pre-censor, the regulatory regime resulted in one network choosing to bleep the word “beef” out of a F.R.I.E.N.D.S. episode about burgers. In short when there is a broad and unreasonable legal obligation to censor, companies will err towards over enthusiastic content removal to avoid being dragged through a lengthy, expensive litigation.
It helps to remember that there are many laws in India that criminalise various kinds of speech. The Indian Penal Code alone covers a range of speech, from sedition to insulting religion. While pre-censorship may be justified in the context of child-pornography, its harmful effects eclipse its benefits in the context of defamation or sedition. Speech flagged as defamation often consists of valuable information about public figures and officials that is important to India’s functioning as a democracy.
For instance, the defamation law was used by climate change expert RK Pachauri against the women who alleged that he had sexually harassed them. Several journalists have at least one defamation case filed against them by a major company or politician that they have investigated.
Major online platforms already err towards censorship even without the legal pressure to over-comply. The UN Special Rapporteur for Freedom of Expression has had to write to Twitter about its censorship of content from Kashmiri journalists from its platform, and the New York Times has reported that Facebook has instructed its moderators that the phrase “Free Kashmir” should be removed whenever reported. Imagine how much worse this problem will get with the obligation for automated removal of unlawful content in place. Valuable political speech will never see the light of day.
The filtering out of such speech would not only violate the individual right to freedom of expression, it would destroy our access to information. If Indian citizens’ access to information is curtailed, democracy is threatened.
Other developments related to the IT Amendment
This proposed amendment did not emerge in a vacuum. The more objectionable ideas that it attempts to codify have emerged from elsewhere in the legal system. For example, forcing companies to incorporate in India echoes the Srikrishna committee’s data protection bill that requires data about Indian citizens to be hosted in India. In combination with the amendment, it would the give the law enforcement ready access to our social media data.
The automated removal of unlawful content may have its roots in hearings before the Supreme Court in Prajwala v. Union of India. The case is about non-consensual pornography online, and the major global online platforms involved have worked with the government to develop a black-box system for filtering content. The question of automated removal of content has come up in this case, and appears to have been accepted by all concerned. There is a possibility that this is the background to automated removal now being proposed for all unlawful content online.
A state-curated version of news
The proposed amendment will hopefully be rejected following public consultation. It displays a tendency to control information in an opaque manner, and to co-opt all the major web based intermediaries to do this. These changes would damage the democracy and the economy.
It is unfortunate that India is choosing to look at big data and algorithms as technology that can control the public sphere and monitor citizens closely. Ours is a constitution that celebrates access to information. Our role in the region and in global politics has been dramatically different from China for this reason.
Although several of our laws regulating speech were drafted to keep a colony in check, many of these have been read narrowly by the Supreme Court, to exclude speech protected by the Indian constitution. This nuance is often lost when arrests are made and hearings in the lower courts take place. It will certainly prove elusive to an algorithm, or to an exhausted manual reviewer who has been instructed to err on the side of placating the government.
When a private company makes errors or censors constitutionally protected speech, there is no mechanism by which one can ask for judicial review. The company itself is not required to offer detailed reasons or an avenue for appeal.
If the government suggests to companies that news of the shelling of civilians in Kashmir, of farmers marching to Mumbai and Delhi to protest, and of deaths resulting from demonetisation is sedition or incitement to violence, the social media platforms, search engines and news aggregators might disable access to all this content. Like Chinese residents, we might find ourselves reading a state-curated version of the news. We would hear nothing about the corruption of politicians or the suffering of our fellow citizens. A public sphere controlled by automation is a finely controlled public sphere. Our democracy would never survive it.
Chinmayi Arun is a Fellow at the Berkman Klein Centre at Harvard University.