As millions of migrant workers made their way from India’s cities back to their villages after the government announced a nation-wide ‘lockdown’, the Solicitor General of India informed the Supreme Court that the exodus was caused by “some fake/misleading news and social media” and sought a direction to prevent “fake and inaccurate reporting” (here). In Maharashtra, an order was passed under Section 144 of the Code of Criminal Procedure prohibiting the dissemination of information on social media that was ‘incorrect or distorted facts’ (here). And Kashmir’s new ‘Media Policy – 2020’ states that “Any fake news or any news inciting hatred or disturbing communal harmony shall be proceeded against under IPC/Cyber Laws” (here).
Phrases such as “incitement” and even “disturbing communal harmony” have a long and well-documented use in Indian law (for better or for worse). However, the above narrated incidents demonstrate a recent trend by the Indian government to try and restrict speech on the ground that it constitutes “fake news” (I use speech in the broadest possible term to include the press, broadcasting and online media). India is not alone, countries such as Singapore and Indonesia have introduced full-blown legislation to restrict “fake news”. In this post, I begin by noting that the term “fake news” suffers from several definitional hurdles that point to deeper structural problems in our media eco-system. I argue that there are several very good reasons why we may want to restrict some forms of misinformation. However, any restriction imposed on speech must comply with the constitutional safeguards set out in Articles 19(1)(a) and 19(2). Examining “fake news” restrictions against the concepts of vagueness, overbreadth and a disproportionate chilling effect, I argue that restrictions on “fake new” that are narrowly tailored enough to be constitutionally compliant are unlikely to be effective in combatting the social harms we associate with “fake news”. I conclude by advocating a heterogeneous approach to combat the issue of “fake news”.
A few caveats. First, because India does not yet have a “fake news” legislation, my analysis is necessarily in the abstract (even the Kashmir policy ultimately relies on provisions of the Indian Penal Code for prosecution). This post seeks to evaluate the consequences of restricting “fake news” as a category of speech and I accept that any restrictions imposed by the government may be more nuanced than a blanket restriction on “fake news” (although the signs are not promising). Second, there is a separate but cognate conversation to be had about the role of internet intermediaries in facilitating and restricting “fake news” that is worthy of a separate post and I have not addressed the issue here for the sake of brevity.
Protected Speech and its Limits
Before beginning it pays to recap a few important aspects of free speech regulation in India. While Article 19(1)(a) guarantees citizens the freedom of speech, Article 19(2) allows for “reasonable restrictions” in the interests of inter alia: (i) the sovereignty/integrity of India; (ii) the security of the State; (iii) public order; (iv) decency or morality; (v) defamation; or (vi) incitement to an offence. As we can see, speech in India can be restricted because of its consequences, that it may lead to violence, but also because of the speech’s content – that the meaning conveyed is deemed legally objectionable. The State evidently has an interest in restricting speech that directly leads to violence. However, in the case of obscenity laws or defamation, speech is restricted because of value judgements by the State. Obscene speech does not lead to violence, but the State believes that it leads to an erosion of public morality.
Any restriction on speech must have a proximate connection with a specific head set out in Article 19(2). The government cannot restrict speech merely in the ‘public interest’, or because it is ‘false’, neither of which are heads under Article 19(2). Therefore, if the government wanted to restrict “fake news” it would need to prove that “fake news” either caused harm because of its content (defamation, decency or morality) or that it was inciteful leading to violent consequences (public order, incitement to an offence). Lastly, there is a long line of cases noting that the ‘proximate connection’ means a real and imminent risk of harm arising from the speech and not vague speculation about possible future harms.
Defining “Fake News”
“Fake news” is a term bandied about very loosely nowadays which has resulted in everybody thinking there is consensus about the phenomenon being referred to, but very little certainty as to what content is “fake news” and what content is not. The term has been applied to satire, propaganda, biased reporting, sponsored or promoted content, factually incorrect reporting, entirely fabricated stories, or simply inconvenient truths. The term does not clarify whether it applies to private communications (WhatsApp chats), social media (Facebook), online media (an online-only news organisation) or even traditional print media. In a post Donald Trump era, the term also necessarily carries a derogatory component that is often independent of an objective evaluation of the actual content (Habgood-Coote refers to this as an ‘epistemic slur’). The flip side of this issue is that there is very little certainty about who a “journalist” is today, with citizens receiving news from a wide variety of sources.
The term “fake news” therefore refers to a heterogeneous field of content, some of which have a diverse set of underlying problems. This becomes immediately apparent when we look at the words we used to use to describe this type of content before we acquired began using the umbrella term “fake news”. Inaccurate, false, misleading, biased, sensationalist, propaganda and advertisement are just some of the words to describe what we now call “fake news”. Using an exact term to identify the issue with a piece of content allows us to create targeted and meaningful solutions. For example, the type of regulation needed to regulate factual inaccuracies in a newspaper article is very different from the type of regulation needed to ensure paid advertising can be distinguished from news stories. We need to stop using the umbrella term of “fake news” and begin accurately labelling the specific harms caused by the speech in question.
From a free speech perspective, it is important to note that the distinction between “fake” and “real” is a politically contested one. The more polarised a society, the less likely it is that different parts of a population experience the same political reality. This makes legally regulating the fake/real distinction problematic, as seen in Singapore where the government sent orders to Facebook to “correct” individual posts. The posts alleged that the Singaporean government had illegally influenced investment companies, suppressed whistle-blowers, and rigged elections (here). This is emblematic of how governments can use a restriction on “fake news” to restrict a broad range of criticism and plenty has been written about the Indian government’s efforts to reshape the narrative in Kashmir (here). At the end of the day, governments place a pre-eminent value on self-preservation coupled with a bias towards their own political ideologies, and the ability to determine what is “fake” and what is “real” goes a long way towards silencing opposing viewpoints and homogenising political thought.
That said, there exist more nuanced definitions of “fake news” and a blanket ban on “fake news” is unlikely. For example, the European Union (in non-binding documents) doesn’t use the term “fake news” at all, rather it defines “disinformation” as ‘verifiably false or misleading information which cumulatively is created, presented and disseminated for economic gain or to intentionally deceive the public and may cause public harm intended as threats to democratic, political and policymaking process as well as public goods’. By requiring an evaluation of the intent behind the creation of the information and a (rather soft) the requirement for actual harm the European definitions seems to be less of a burden on free speech. But this also means it is less effective at curbing the spread of misinformation. Users may spread disinformation legitimately believing it to be true, and how does one assess when a ‘policymaking’ process is harmed? It also does not regulate other harms we associate with “fake news” such as media bias. It is worth keeping these considerations in mind as we consider the harms arising from “fake news” or “disinformation”.
The Argument for Regulating “Fake News”
There are two primary reasons why a government may legitimately wish to regulate “fake news”. First, misinformation or fabricated stories may directly result in violence, either through information about specific individuals or more broadly stoking pre-existing fissures in society (e.g. race or religion). India has already witnessed a string of violent incidents that investigative authorities have noted were either caused by or aggravated due to the spread of disinformation. Second, disinformation can interfere with the electoral process by misinforming voters about candidates’ political opinions, track records, previous misdeeds, and positions on important social issues. An uninformed voter is unable to vote for a candidate that best represents their interests, directly undermining the legitimacy of the electoral process and the resultant government. To make matters worse, disinformation is likely to disproportionately effect voters who do not have access to multiple sources of information (the poor and marginalised).
As Ari Waldman notes, in the long run disinformation can fundamentally alter public discourse by creating false equivalencies, particularly where political leaders spread disinformation. If a factually incorrect statement by a political leader must be accorded equal weight as the truth in the name of journalistic neutrality, a society may waste valuable public time and energy debating false stories (e.g. see the amount of time the U.S. media spends merely correcting President Trump’s demonstrably false statements). Over time, sustained bias or propagandistic reporting can harden political bias, causing citizens to select media sources that merely confirm their existing notions of true or false, further increasing polarisation.
To be clear, traditional justifications for the freedom of speech such as Mill’s argument that more debate over ideas ultimately lead to the truth (culminating with Justice Holmes famous ‘marketplace of ideas’ analogy) are not strictly applicable to “fake news”. To argue that we should produce even more “true news” (counter-speech) rather than restrict “fake news” is problematic for two reasons. First, it doesn’t work and inaction can lead to the type of cyclical harms referred to above (a recent MIT study found that false stories diffused further and faster than true stories on Twitter in all categories of information (here)– in other words, the truth does not rise to the top).
More fundamentally, Mill’s theory and subsequent adaptations are premised on having more ideas and counter-ideas, not facts and “counter-facts”. Facts and opinions are very distinct in their nature but also share an important relationship when it comes to free speech. While there may be an infinite number of ideas or opinions about a fact, a fact is singular and objective. For example, whether there are Chinese soldiers inside what India considers its territory is a question of fact. The soldiers are either there, or they are not. What India should about the Chinese soldiers and its own territorial claims is a question of opinion with many possible answers. However, it is also important to acknowledge that “factual truth informs political thought” – in other words people’s opinions have a factual basis. For people to have meaningful opinions about the world, and for those opinions to interact with each other in meaningful debate, there must be an agreed upon factual basis. It makes sense not to restrict ideas and opinions because they open new avenues of thinking, however an agreed upon baseline of facts is fundamental to an informed debate about ideas. Returning to our example, how can we debate whether the government’s foreign policy was a success if we do not know whether the Chinese soldiers were in Indian territory or not? Therefore, a freedom of speech justification created to ensure free debate may actually favour some restrictions on “fake news” by ensuring a common baseline of facts.
Legally Restricting “Fake News”
Recall that under the constitutional scheme, all speech is free other than that which the government restricts because it is has a proximate nexus with the specific harms identified by the heads of Article 19(2) (public order, defamation etc.). Although the Indian Supreme Court has never explicitly dealt with the question of whether factually false statements are protected under the Constitution (the U.S. Supreme Court in Alvarez explicitly held such statements to be protected under the U.S. Constitution) factual inaccuracy is not a ground to restrict free speech under Article 19(2). Therefore, it is reasonable to assume that if the government wanted to restrict speech it classified as “fake news”, it would have to argue that the speech was either defamatory, would lead to a breach of public order, or amounted to an incitement to an offence. What amounts to a valid restriction under these heads of 19(2) have been widely discussed elsewhere on this blog and I do not intend to rehash them. Rather I argue that restrictions on “fake news” are likely to fall foul of two principles: (i) vagueness, (ii) overbreadth leading to a disproportionate chilling effect on speech.
Vagueness: A law is unconstitutionally vague if ordinary citizens cannot determine whether they have broken it or not. Imagine a law which restricted citizens from “honking too much” at traffic signals. How does a citizen determine how much is “too much”? The law does not provide a fair warning to citizens as to whether their actions will break the law or not. Vague laws also grant officials a large amount of discretion as to when a law has been breached or not (would you trust a police-officer to tell you when you are honking too much?).
Restricting speech on the ground that it is “fake news” would likely lead to widespread confusion about what kind of speech the government was restricting. Recall the broad range of content that comes under the banner of “fake news” – from satire to inaccurate reporting to fabricated stories. More nuanced definitions (such as that proposed by the European Union) may remedy this issue but are still unwieldy. The line between advocacy and deception is often imperceptible, with compelling arguments often cherry-picking or manipulating facts. Further, terms such as ‘threats to the democratic, political and policymaking process’ are so wide that they substantially increase the risk of the government selectively prosecuting speakers with unfavourable opinions. Kashmir is a living example of where free speech has been left entirely to the whims of the executive, and it has led to the denial of internet services, the arbitrary blocking of websites, and the persecution of journalists.
Overbreadth: A restriction on speech is “overbroad” when it restricts both the speech that the government can legally restrict (e.g. hate speech) but also goes on to restrict speech other speech that is constitutionally protected (e.g. dissent). In Shreya Singhal, the Supreme Court was called to adjudicate upon the constitutionality of Section 66A of the Information Technology Act, which criminalised speech that was “grossly offensive”. The court noted that any citizen may advocate a view on governmental, literary, or scientific issues that may be unpalatable or even “grossly offensive” to other citizens. While some speech may be justifiably restricted, not all speech that was “grossly offensive” rose to the level where there was an imminent risk to public order or incitement. This meant that, by using the term “grossly offensive” Section 66A also ultimately criminalised constitutionally protected speech. The court observed:
“Section 66-A purports to authorise the imposition of restrictions on the fundamental right contained in Article 19(1)(a) in language wide enough to cover restrictions both within and without the limits of constitutionally permissible legislative action […] It must therefore, be held to be wholly unconstitutional and void.”
The term “grossly offensive” was so broad that it restricted some speech justifiably, but also other speech unjustifiably. In Shreya Singhal the court distinguished between discussion, advocacy, and incitement. It held that the first two formed the heart of constitutionally protected speech while inciteful speech may be legally restricted. Simply put, restrictions on speech that restrict inciteful speech and are broad enough to go on and restrict discussion and advocacy are unconstitutional.
If the government were to restrict “fake news” (a very broad category of speech), it would also likely restrict a vast amount of constitutionally protected discussion and advocacy. At its bluntest, satire, parody, and anti-government reporting could fall under the ambit of “fake news”. More critically, a restriction on “fake news” would create the risk of liability for journalists and media houses, that a single inaccurate factual assertion could lead to censorship or punitive action. For example, Kashmir’s new ‘Media Policy’ flat out dis-empanels journalists for publishing “fake news”. A news-report that exposed governmental overreach but contained a minor factual error or irregularity could be termed “fake news”, robbing readers of valuable information and robbing journalists of their livelihood.
Where restrictions on speech are vague, overbroad, and punitive, they create a chilling effect on speakers. Content creators, journalists, publishers, media houses and ordinary citizen are likely to ‘err on the side of caution’ and simply not speak rather than run the risk of their speech being adjudged as a crime. To avoid this chilling effect, the law of defamation incorporates what is known as the “actual malice” standard. It states that unless a speaker’s statement evidences ‘actual malice or a reckless disregard for the truth’, the statement cannot be considered defamatory. The logic is simple, with the vast amount of free-flowing debate in society a few inaccuracies are bound to crop up. Rather than aggressively prosecute these minor inaccuracies and frighten all other speakers, the law states that where these inaccuracies do not possess any ‘actual malice’ they are exempt from prosecution. In Rajagopal the Indian Supreme Court adopted the ‘actual malice’ standard in civil defamation cases and recently, the Madras High Court has adopted the standard in the context of criminal defamation (here). Defamation requires an alleged injury to reputation, something not all “fake news” is likely to cause. However, the rationale that minor inaccuracies should not lead to punitive action which has a chilling effect on speech should certainly apply, and any restrictions on “fake news” should incorporate the ‘actual malice’ standard.
The Government and the Truth
One last point may be made before concluding. In Alvarez, the U.S. Supreme Court struck down a statute which punished persons for falsely claiming to have been awarded military medals. In striking down the law, the court noted that that it was wary of the government claiming ‘broad censorial power to regulate falsehoods’, the mere existence of which would have a chilling effect on speech. To be clear, the government does regulate falsehoods in certain cases, most notably in the areas of defamation, consumer protection and fraud, and perjury. This is accepted because, the restrictions on speech are limited to the contexts where they are especially likely to cause harm and are actionable only where they cause actual harm to identifiable individuals. A restriction on “fake news” is much broader. A good news eco-system is more akin to a public good like a clean environment and “fake news” often attacks issues more than people – where it does attack people, the remedy of defamation is always open. Similarly, where the “fake news” is inciteful and capable of causing violence, India already possess a host of anachronistic laws that can be used to restrict and prosecute speech capable of causing violence. Without this proximate connection to real threats, regulating “fake news” creates the spectre of a government truth, something any democracy should be wary off.
The phenomenon often branded as “fake news” can lead to a diverse set of harms ranging from violence, damage to the electoral system and increased political polarisation. However, when examined in light of India’s freedom of speech jurisprudence, restricting “fake news” is constitutionally fraught. Above all, it is important to remember that the government cannot restrict speech merely because it is “false” or “inaccurate”. Briefly, any restriction on “fake news” would need to (i) show some real, imminent and identifiable harm; (ii) the harm would need to be a constitutionally recognised ground to restrict free speech under Article 19(2); (iii) consider questions of medium, time and impact of the “fake news”; (iv) incorporate the ‘actual malice’ standard; and (vi) be the least restrictive measure at the government’s disposal. This is not to say that it is impossible to draft legislation regulating “fake new” (several countries have already done it). There may be real value in pursuing narrow restrictions to reduce the effects of active disinformation campaigns during especially sensitive times (e.g. the run-up to elections). However, the phenomena of “fake news” rarely appears in such convenient forms and speech restrictions that are constitutionally compliant are unlikely to address the other diverse and systemic harms that “fake news” causes.
I began by noting that “fake news” is an amalgam term that houses several underlying issues we associate with our media-ecosystem. Understanding the underlying issues can lead to targeted solutions that bolster media literacy amongst the population and reduce the economic and political incentives associated with fabricated or propagandistic stories. Examples include ensuring a competitive media industry, disclosures requirements on social media sites about promoted content, investment in a truly independent state broadcaster, media literacy education in schools, and ultimately targeted legal interventions where constitutionally compliant speech restrictions may be meaningfully enforced. Merely sounding alarm bells at the rise of “fake news” opens the door to restrictions on speech and government censorship.
This post was largely in response to Kashmir’s Media Policy, which along with the Solicitor General’s statements represent a worrying trend by the government to adopt the clumsy but incredibly dangerous fake/real distinction to assert a dominant, government controlled narrative. In the future, I am sure there will arise more concrete examples that lend themselves to more detailed critique. The government’s position has long been that speech in Kashmir rejects the paradigm of the Indian State itself, representing a unique threat to the integrity of India. However, where the government seeks to preserve power through a stranglehold on truth itself, we have to ask ourselves whether the means employed to preserve the State have overridden what the State once stood for.