New Delhi: In a piece published recently by the Guardian, Julie Posetti and Kalina Bontcheva wrote that the 8.5m tweets directed at journalist Rana Ayyub in 27 months marked her out as one of the most brutally targeted journalists in the world and exposed Twitter’s failure to address online violence. 

Ayyub, an independent journalist, Washington Post columnist, and a critic of Prime Minister Narendra Modi and the ruling Bharatiya Janata Party, has used her social media heft and the global attention she receives to highlight the plight of Indian Muslims and the arrests of journalists in India. She is accused of money laundering and tax fraud related to her crowdfunding campaign to help those affected by the Covid-19 pandemic in 2020.  Ayyub has denied any wrongdoing, calling the allegations “baseless.”

The tweets were examined as part of an international research project aimed at developing an early warning system for gender-based online violence against women journalists, in light of the mounting concerns of online abuse and harassment leading to offline attacks, wrote Posetti, deputy vice president and the global director of research at the Washington DC-based International Center for Journalists (ICFJ), and Bontcheva, a professor at the University of Sheffield in the UK.

In an interview with Article 14, Posetti delved into why they picked Ayyub for the study, the nature of coordinated attacks, how online violence enables political repression, and why the tech companies needed to fight the abuse levelled at women journalists in many countries, as well as the state-issued orders of censorship on its platform. 

On 26 June 2022, complying with an Indian government order, Twitter withheld Ayyub’s tweet from 9 April 2021 on the Gyanvapi mosque after a local judge allowed an Archeological Survey of India (ASI) survey of the religious site in Varanasi. That same day, Alt-News co-founder Mohammed Zubair tweeted that Twitter had withheld his tweet from 22 February on a right-wing Hindu leader. The next day, Zubair was booked for outraging religious sentiments and arrested by the Delhi Police for a tweet from 2018 with a photo from a Hindi comedy film. In the subsequent weeks, Zubair was booked in six other cases in Uttar Pradesh. 

On Twitter moving the Karnataka High Court to challenge the Indian government’s orders to take down content, Possetti said it was a good move. Still, social media platforms had to take responsibility for what was happening on their platforms, prioritising people and fundamental rights over profit and engagement. 

“What I found dichotomous is that you can have an individual journalist like Rana Ayyub threatened with rape, murder and all sorts of violations on a daily basis and this isn't acted against but what is acted against is a series of what appear to be innocuous tweets and those the state seeks to repress,” she said. 

What made you think of Rana Ayyub as your subject?

I've been following and researching and reporting on Rana’s case for several years now, in fact since I think about 2017, which is when I first became aware of her situation. Based on our international research, it is very much an emblematic case within India but also internationally. The idea for our studies so far—for these novel big data case studies—is to identify not necessarily high profile but examples of women journalists who are prolifically targeted so that we can study and therefore better understand the nature of the abuse in real-time and also to then make recommendations about how better to address and respond to those challenges. Rana’s situation marks her out as one of the most prolifically abused women journalists in the world. She was therefore one of the five women who we have studied for this report—a series of reports—that we have produced on commission from the UK government. 

Rana Ayyub has been called a divisive figure, with scores of people who support her and those who don’t agree with her point of view. In picking Rana, do you politicise what is also an academic study? 

I don't think so. Journalists are by their nature inclined to be accused of being divisive because they report critically on issues connecting to power. It is therefore impossible to avoid politics when analysing or selecting targets when you study journalists. I am a  journalist myself and have been an academic for a long time as well. We are not studying politics per se, we are studying the methods and forms of attack on a journalist and on a series of journalists all of whom have been framed as divisive and in some cases accused of being criminals. Rana is not alone in that. Among the women we study, Maria Ressa is another one in the Philippines. They are targeted because of what they say and what they report but it is not our job to determine whether or not what they say is divisive or not. It is our job to assess the kinds of attacks that they attract. I think it is deeply problematic to suggest that what they say is in part responsible for the abuse and the online violence they experience. It’s a form of victim blaming. 

You write that the abuse and harassment women experience on social media platforms is online violence. What is online violence?

In the same way that with domestic violence—as definitions have evolved over the past two or three decades—we know that domestic violence doesn't include just physical assault, it includes economic violence, which is the ways in which women are exposed to poverty through abuse by their partners, whether that is restricting access to the financial resource or being prevented from earning income. We also see psychological violence referred to with reference to domestic violence. And we now refer to digital violence in domestic violence cases.

We chose the term online violence because online harassment and abuse are the terms that are frequently used to describe what happens on social media platforms and other digital devices and areas of online communication. And violence is used very deliberately because violence is not just kinetic, physical violence.

Violence takes all these other forms. In the case of journalists, the effect on these journalists includes very serious psychological injury in some cases. It also escalates the risk of physical violence. We have various cases of women journalists who have been abused, attacked and harassed online, experiencing offline harm they attribute to the climate of abuse they experience online. 

It also means a variety of manifestations, so online violence includes threats of physical violence whether it is rape threats, threats of sexual violence in other forms, threats of physical violence including murder, threats of harm against immediate family members particularly children, which is a phenomenon we have tracked. Additionally, it also includes digital security attacks. We see women journalists frequently being doxxed, as a way of escalating the risks they face offline, to increase their exposure to physical risk, and we see these things happening with impunity. They are very much in parallel with the traditional threats that journalists experience in a physical context. 

You write that there is a symbiotic relationship between online violence and political repression. (The former chills press freedom, and creates a more permissive environment for the latter.) Do you see this as the future, the immediate future?

We certainly see it happening now. We saw it in (Daphne) Caruana Galizia’s case and evidence that has been accepted by the public inquiry in Malta for murder. We see it in Maria Ressa’s case where there was first an attempt to seed public hatred and public acceptance of her persecution and prosecution which became an eventuality and ultimately her conviction. And we certainly see it in Rana’s case too. A lot of the online violence that she experiences has elements of the organisation about them with the same kinds of threats used against her that we see in Maria Ressa’s case—constantly accusing her of being a criminal although there are no charges pending against her at this point. Those are the sorts of manifestations, hashtags like #arrestranaayyub, which also occurred in Maria Ressa’s case and others as well. This use of online platforms to call for offline persecution is all designed ultimately to chill those journalists’ speech,  but also curtail their critical reporting. 

What are the legal efforts to arrest this abuse? 

There are moves to legislate, to incorporate, sex and gender-based hate speech in hate speech laws for example in various countries, particularly in Europe. Those efforts are designed to ensure that women who might not be able to claim protection under existing laws towards discrimination, whether that be race-based discrimination or religious-based discrimination, can also claim protection under the laws on the basis of threats and abuse experienced due to their sex or their gender. I think that's an important development but we also see legal and regulatory responses to the ways in which the platforms have failed to address these problems in an effective way or have chosen to abrogate the responsibilities they have under U.N. principles. 

One of the key issues is that the dominant platforms are still US-based tech companies. There is a really important role for US regulators to play in ensuring accountability from the platforms and that accountability needs to be addressed in a way that allows protections globally.

If someone is being abused in Hindi or in Tagalog in the Philippines, or whatever the situation might be, is able to get support and redress from the platforms knowing they have the capacity to respond to abuse in those languages as well. To date, there has been a dominant kind of approach that has focused on the English language which is to the detriment of other women around the world who are abused in multiple different languages on a daily basis. Those issues are important in terms of social justice. 

And all of this has to be done with a view to trying to prevent broader threats to freedoms of expression. We, unfortunately, see states and governments use such protective laws against journalists, for example. We've seen that with fake news laws, which involve laws that are theoretically there too —in inverted commas—protect people from disinformation—but ultimately are used against journalists perhaps for reporting critically on pandemic responses, for example. It is a complex area but we are seeing a range of different responses which need to be taken seriously and treated with urgency. 

You write that the harassment Ayyub receives is an intersection of misogyny, anti-press sentiment and religious bigotry. Why those three things? 

She is a woman, she is a journalist and she is identifiably Muslim and that is borne out in the data. We see her being threatened with rape and violated in sexual ways which are misogynistic and hateful, we see her being abused because she is a Muslim and falsely accused of being a jihadi, and subjected to all manner of other abuse based on the assumption that her faith makes her a traitor to India. We see much anti-press sentiment which is a global phenomenon where journalists, in particular, are targeted and abused, partly due to a politically generated loathing of journalists often triggered by political actors who would prefer not to be held to account by critical independent journalists. That is everybody from (Donald) Trump to (Jair) Bolsonaro to evidence that we see in the Indian context of supporters of prime minister Modi. 

Are women in other countries targeted because of their identity? 

Absolutely. In all of the case studies that we have done so far, we see what is referred to as intersectional identity, where the woman journalist is also a target  of  discriminatory attacks —whether it is her faith, whether it is her race, her ethnicity—those factors tend to increase the risk of exposure to online violence and make not just make the impacts worse but deliver some of the worst instances of online violence. We see it with Ghada Oueiss who is a journalist with Al Jazeera who is based in Qatar and is Lebanese and she is a Christian working in a Muslim-dominated part of the world. She is frequently targeted not just because of her sex but because of her Christian faith and the fact that she is living a lifestyle that is condemned by some in that region as being permissive. because she dares to have a social life and do all things that women do all over the world on a daily basis in their public lives and their private lives.

Maria Ressa is also targeted because of her skin condition. So, you see, disabilities come into play as well. She has a skin condition and that is usually mocked and she is called a whole range of things including using terms for male genitalia, comparing scrotum skin to her face.

There is horrible, just horrible, dehumanising intersectional abuse levelled at women journalists all around the world, so much more vile than the abuse we see hurled at men. 

You point out the speed at which abuses are hurled at Rana Ayyub after she posts. What does the speed tell us and what is a coordinated attack? 

The speed at which abuse is hurled at Rana is highly unusual. It comes at her thicker and faster than any of the other cases we studied. We know from other research that when you see abuse coming at such speed and with such intensity—within seconds every time she tweets—such rapid-fire abuse can be a pointer towards coordinated attacks. By orchestrated or coordinated abuse we mean it could be centrally coordinated, or deliberately coordinated. We have seen various instances of state-linked and dark PR firm-led attacks on journalists internationally. I’m not saying that's the case in Rana’s situation. We are still processing the data with regard to that. But we can see patterns that are associated with orchestration. In other words, to see so many tweets that threaten and abuse, at such a rapid-fire pace is highly unusual. And the fact that we are using systems of analysis that are incredibly conservative so we know we are missing a whole lot of abuse. But the computer scientists that we work with use an academic level standard to try to ensure that the abuse they are identifying is indeed obvious, very clear abuse. What that says is that this is a conservative estimate of the abuse and the speed at which it arrives rather than an overstatement by any stretch.

There are various forms of coordination. There is what I mentioned before—the kind of very deliberate orchestrated coordination. We also see pile-ons that are more organic. They can be triggered by a misogynistic group.  For example, we see this with women journalists in the US where there is a behind-the-scenes attempt to inflame and escalate a pile-on that is in itself a form of orchestration.

You have studied over 8.5 million tweets targeting Rana Ayyub. What are the resources and timeline for a project like this? 

We began collecting tweets directed at Rana in December 2019, and we work with a team of academic computer scientists at the University of Sheffield, who have data scientists as part of their team. The computational analysis is done by them and that's machine learning assisted. These are massive data sets so we are not manually extracting tweets and examining them in isolation. We are pouring these tweets through an automatic system that is examining the content of those tweets for the kind of linguistic markers for abuse that have been identified through successive studies including some common Hindi terms of abuse that have been identified by Hindi-speaking research scientists in Rana’s case. The computer scientists are using machine-learning assistants to apply what is called natural language processing, which is an automated way to analyse the text with large data sets and then we are able to dig in and further analyse in isolation the subsets of that data. It is incredibly labour intensive and expensive and involves a hell of a lot of skill and then we match that with the kind of qualitative and contextual research that we do in the research team at the international centre for journalism which is more like I suppose in journalistic terms high-end investigative reporting than it is a machine learning based computer science natural language processing. But we then blend those techniques—the qualitative and quantitative—to be able to tell the story that is associated with the data. 

You write that there are chilling similarities between Rana Ayyub’s case and those of Gauri Lankesh and Maltese journalist Caruana Galizia. How so?

In Caruana Galizia’s case in Malta, she was the target—I say that specifically because she was not actually on social media—but social media, particularly Facebook,  in Malta, was awash with threats and abuse targeting her. And so, even though she was not herself logging into Twitter and Facebook and finding the abuse directed at her in her inbox or on her profile because they didn't exist, this was evident in chat groups, it was evident in various communications and on various pages on Facebook. Daphne (Caruana Galizia) was threatened with rape, she was threatened with murder.

She was called a witch, she was called a criminal, and she was abused and harassed with all manner of heinous terms that included the phrase presstitute, which is the one we see used against Rana Ayyub, Maria Ressa. What is chilling is the way in which the threat was made, the ferocity of those threats, the way they created this permissive culture for abuse and the targeting of the women who were at the centre of these campaigns.

In the case of Gauri Lankesh, of course, Gauri was harassed online, she was harassed specifically because she translated Rana’s book and she was harassed in a way that led Rana to talk to her about her experiences of online violence as we refer to it now. This was in the days even prior to her murder so there is a direct connection between those two women and their experiences are somewhat similar in terms of the threats and abuse they received. 

You write that Twitter has guidelines against harassment and protecting free speech, but these are flouted. What should Twitter be doing?

One of the problems and this is why Twitter has gone to court in India is that states attempt to force the platforms to comply with particular laws that are political in origin—to do with censorship rather than to do with the protection of individuals who are under international human rights law should be protected against hate speech, for example. Journalists are a subset of individuals along with human rights defenders who have particular protections with regard to freedom of expression and the right to safely practice journalism. That balance is out of whack. Twitter has, whether or not it is in response to concerns about censorship of journalists in India, whether that is one of the reasons for them going to court, we don't know. The fact they are going to court is a positive sign because they are trying to presumably reinforce the rights of the users, but it is a delicate balance as I said before and one of the things they need to do differently is not just leaving it at this point to make legal representations to allow them to do what their policies require, but that they also actively seek to protect women journalists under attack. For far too long, they have not understood or have not acted to appropriately defend and protect women journalists. One of the problems with that is that there is limited capacity for women journalists under attack to escalate with Twitter, to get interventions, to censor and de-platform people who are routinely hurling abuses. It's not just Twitter. It’s Instagram as well.

In Rana’s case as well, she has a hell of a lot of abuse coming from Instagram, particularly via the private messaging functions of that platform. It is about ensuring that people who have protected the freedom of expression right under international law, are not just talking about unfettered speech. Somebody who is an abuser is not protected by freedom of expression laws. They are not designed to support hate speech and abuse.

With social media platforms, let's say Twitter dealing with journalists, there need to be much better and more effective ways to escalate complaints and to get action, to have a human point of contact in the loop, particularly in the context of offline threats with which journalists are being targeted are particularly significant. That's what we see in the case of India.

Is Twitter taking the Indian government to court a welcome step? Has it done so in other countries?

Twitter occasionally pushes back legally. In India, it is a special case because up until recently, till last year, they were able to operate under policies that identified journalists as people who need to have their reporting on the platform appropriately protected but under the current law in India, there are serious repercussions for the representatives of Twitter or any other company in the country if they are not enacting the censorship instructions that the government issues. This is extremely problematic. I welcome any attempt to challenge the necessity to abide by state instructions, whichever country's context, to censor journalists and to censor critical commentary that by any standard wouldn't ordinarily be perceived to be threatening or problematic. What I found dichotomous is that you can have an individual journalist like Rana Ayyub threatened with rape, murder and all sorts of violations on a daily basis and this isn't acted against but what is acted against is a series of what appear to be quite innocuous tweets and those the state seeks to repress. I think Twitter is right to challenge that and try to draw attention to the problematic nature of such laws. But they also have to as a corporate entity take responsibility for what happens on their platform and make decisions that are based on their own ethics and principles of international human rights. 

In your Guardian piece, you quoted Rana Ayyub as saying “the trolls and the state have realised that these social media platforms are dependent on them”, and they are prioritising profit over people. When and why did this happen? 

This is something that we have observed and has written about broadly, with reference to social media platforms, responding to online violence against women journalists in various U.N. reports. The reason we have concluded is that the platforms are prioritising profit over the implementation of human rights protections—this is something that for the last decade has been debated—and is evident in the lack of appropriate, effective responses. The vast wealth of these companies, which are only addressing these concerns when they are dragged kicking and screaming to do so,  is what leads to the conclusion that profit, which is based largely on engagement which is what feeds the algorithms, is the dominant purpose of their operations. It is not, despite their statements about the altruistic values of connecting humans to another—that may have been the original design of the platforms to a certain extent—but the ways in which these platform has evolved, the power which they have amassed, the huge profits which they have acquired have unfortunately not been matched with progress with regards to developing the technology and taming the algorithms such that they chose to protect and defend human rights. That includes press freedom, the protection of women from threats of rape and other forms of abuse and it also includes ensuring that powerful people can be held to account. These are all factors that the new media—the social media—have in their development failed to accommodate. As we saw in the case of journalism, over 200 years of development that such values are supposed to be at the core of practice and in some contexts regulated or self-regulated. We don't see that with social media.  And I think that’s a really significant challenge. And that is why we are finally seeing after over 10 years of debates or more, states that would otherwise not want to address these problems, particularly the US, starting to try and deal with the free-for-all that section 230 (Communications Decency Act) in the US creates—where the platforms are able to operate without any responsibility for the content that they enable and amplify.

Does Twitter depend on trolls? 

It goes back to engagement. I wouldn't necessarily put it exactly like that, but they are dependent on engagement to ensure that eyeballs and the data of the eyeballs are able to be used for the benefit of advertisers, so there is a commodification of engagement and that engagement is tuned to privilege abuse and hate and anger and rage over fair and moderate debate. The algorithms favour hate and fear, over what might be fair and reasoned, and that is the connection between troll armies or individual hatemongers and profit. 

Social media platforms were supposed to facilitate the exchange of ideas and bring us together, not tear us apart. What happened? If it does bring out the worst in us, how much can we legislate or regulate?

There was certainly a  period in the history of social media’s development where democratisation and collaboration and communal kind of conversation of a positive kind felt like the core values of those platforms, but that time has long since passed. I would have said things like that in 2010. I look back and say how naive was I? Maria Ressa says similar things. This concept of social media for social good—still can be used for those purposes but it is extremely difficult now that most of these problems have been weaponized for nefarious purposes, whether it is political actors, states, corporate actors, or criminal organisations. It's at scale very difficult to deal with. And because we see a situation where algorithms are unregulated, and algorithms favour the hate, the argument the anger, the rage, the fear, all of those high-end emotions that tend to generate engagement and are therefore prioritised. It’s the profit factor. But it is also the fact that profit was prioritised over what some users, many users, originally perceived to be these kinds of altruistic goals of these platforms in terms of democratising and cross-border connections and all of the really positive things that they demonstrated that they could be used for in the beginning have unfortunately been muted if not derailed by these other realities. If you look at some of the US legislation being debated at the moment, all bills that are being developed, focus on regulating algorithms as a way of trying to address some of these concerns. We became a commodity. These were not non-profit platforms. It is not the origins of the internet that we are talking about here which did have that kind of ethos at its heart. 

We see Rana Ayyub posting screenshots of the graphic abuse and threats she receives and tagging police handles, but they don't always respond.  How does it compare to other countries? 

This is my analysis based on global comparisons rather than any deep expertise in India.  I realise there are complexities, particularly at the state level. We see all around the world, failures of policing when it comes to understanding these problems when it comes to understanding how to investigate digital hate crimes. There are problems with capacity, training, and culture, and by culture, I mean systemic culture.  The pervasive idea among many international policing entities is that if it's not online, it's not real. If it's online, it's not actionable, none of which is true. There are various legal instruments at the international level that say human rights offline must be upheld online and that should trickle down in the domestic context.

It is shocking that a journalist can be threatened with rape involving implements and graphic descriptions of violation and that no action is taken,  but India is not alone, and we have seen examples in many other countries. And in some contexts, there is not just an absence of action but an absence of a political will. 

Women have lower social status or lower legal status and so this is extremely problematic. There needs to be an understanding and an acceptance of the harm that happens through online violence is serious, and real even if it is in the form of severe psychological injury, or hospitalisation that can come with that, but also that there is now a correlation between online violence and offline harm. It needs to be understood. We are in the process of developing this early warning system because we recognise that signals that happen online can be indicators of offline events. That is something that I think the police everywhere need to understand. In a country like India, where there is now evidence of a connection between online threats and offline attacks, which have happened in the past couple of years, then the propensity for physical or sexual violence against a woman is high to start with, against a woman journalist it takes on characteristics of impunity, for crimes against journalists. Therefore, the police must take this very seriously and must act, recognising that a woman journalist operating on a platform in a journalistic capacity needs to be seen as a target, who demands protection rather than prosecution and persecution.

Are women journalists in other countries also at the receiving end of the extremely horrendous and vile threats sent to Rana Ayyub?

We do see them in places like Mexico, and we do see them in the Philippines where Maria Ressa has been threatened with being raped to death. They are graphic, they are disgusting, but we do see them internationally. Misogyny is a big feature of these attacks on women journalists. And that's universal. 

In Rana’s case, they do seem to be particularly vile and just intense in terms of the speech with which they come but also prolific. Just the spikes we see in abuse. It really is horrendous.

What is this about? It is blending misogyny, and religious bigotry and attempts to silence women journalists whether it is because of misogynistic bots or whether it is linked to state-linked attacks. These are very serious problems. They cannot afford to be normalised.  So many women (journalists) we have interviewed over the years, we have interviewed over 180, are saying things like they felt it has just become so normal, that they didn't even feel safe to raise it with their employers or safe to take it to the police because they felt it wouldn't be treated with seriousness or it wouldn't be viewed as a problem. That is an appalling state of affairs. We have to respond with shock because it is shocking and do something about it. 

Is there any country where this a good response? 

No country that I can say has been without its problems. Even in the UK, one of our research subjects and investigative reporter with the BBC, Marianna Spring, has had an awful time with the police trying to get action. They didn't know what to do with the threats that came at her on social media. They didn't understand how social media platforms work. One of the problems is that there will be sections of a police service in a developed western liberal democracy where this issue is high on the agenda. There will be—maybe at the national level and in one unit of the police—there will be good understanding and a genuine attempt to address this problem as part of a national plan of action to protect journalists, for example. But that isn't necessarily dispersed. When you have journalists attacked in regional areas, journalists attacked in a different state, depending on the nature of the political system in the country, and the way in which policing works, you won't get a universal understanding even in a country that might be doing better than others.  This is a problem that we have to grapple with and it comes down to education and training and policy changes and this is right through from the judiciary to the cops on the beat. 

In India, investigators don't know the difference between archived links and screenshots. The lack of expertise is some part of the problem. 

It is definitely some part of it and again we see that internationally. In one eastern European country, for example, in what was called the high-tech crime unit, the journalist was asked to download the abuse onto a CD-Rom. Another, in a Western European country, was asked to take screengrabs of the abuse and to print it out and post them because they didn't know how to access the platform. It comes down to prioritisation, capacity training and so on. But there is a political element too because policy governs this. If policy downplays this as a threat, then it’s not going to be taken seriously. There is a matter of political will as well as institutional training and capacity and capability. 

You wrote—“How can Twitter let this stand? Even if its staff face criminal liability for non-compliance with Indian government orders, these developments represent a dangerous pattern of concession to state censorship of critical independent journalists.” But the safety of employees is a very serious consideration, especially in authoritarian countries. 

Absolutely, considerations,  But considerations would ultimately come down to this: do you stay in this country? Do you continue to offer a service in a country as much as it might be profitable and valuable, even for the users? These are questions they need to be contemplating. They have to be concerned about the safety of their individuals. That might require them to consider not having individuals in the country which has implications for their business operation. State-ordered censorship of a journalist or journalists on Twitter cannot be accommodated. It is a very dangerous concession in what is an environment that can only worsen once such a concession has been made. We didn't write that lightly. We recognised that the safety and care of their staff are vitally important, but it has to be considered in the context of what is at stake in terms of fundamental rights more broadly and of course they need to take measures to protect them, but making concessions to a state of this kind, we would argue, is a step too far. They have made a decision to tackle this legally, which is one way to go. And that is a good action. We’ll just have to wait and see what the outcome is. 

What do you do when these platforms like Twitter are also the place where people can still say something, and exchange different kinds of news and information especially important when the mainstream media is pro-government.

There is a line in the sand for each of these countries when it comes to despotic states. What is the cost of operating in countries where there is no freedom of expression? If you are a platform fundamentally designed to facilitate freedom of expression. These are big questions. The point we are making is that you can't simply say it's too difficult and we have to just sit back and take it and censor as many journalists as we are ordered to. That's not going to work. Long term it just creates an environment where the censorship just gets more and more prolific and more and more damaging. 

(Betwa Sharma is managing editor of Article 14.)