On razor’s edge
As South Asia struggles with real-world consequences of misinformation and hate speech on social media, the need for auditing of algorithms for bias, punitive measures, and bystander intervention has become evident
Social media in South Asia isn't just a platform for selfies and cat videos; it's now a volatile cocktail of misinformation, hate speech, and violence, with real-world consequences dancing on the digital razor's edge. The recent spread of misinformation and fake news regarding the unfortunate RG Kar case once again makes the social media questionable. A thorough scrutiny of social media is warranted. A conservative estimate for the total number of social media users in South Asia, as of the latest update, is around 961 million. With such vast number of people accessing these social media platforms, the key questions before policy-makers on social media are: how do social media algorithms, designed for engagement, contribute to the spread of hate speech and misinformation; what role do anonymous accounts and bots play in amplifying harmful narratives and inciting violence on social media platforms; and what are the specific ways in which social media platforms exploit religious, ethnic, and caste sensitivities in the region?
Most importantly, what is the role of governments and civil society organisations in holding social media platforms accountable for the harmful content they host? There are two other important questions, always debated in the region: first, how can we balance the need to curb online hate and misinformation with upholding freedom of expression and democratic values? Second, what are the long-term consequences of exposure to hate speech and misinformation on social media for individuals and communities?
There are so far no convincing answers to these questions, which is affecting millions of social media users in the region. Social media platforms, for their political biases towards the governments in power to generate revenue, are surprisingly silent in investing on these questions.
The common pattern of creating social tension through these platforms in South Asia entails sharing misinformation and fake news, trolling and abusing religious minorities, using derogatory language and hate symbols, organising online campaigns to incite violence, sharing caste-based slurs and discriminatory content, inciting violence against ethnic and religious minorities, spreading blasphemy accusations, and inciting violence against journalists and activists.
Data tells a grim story: hate speech spews like venom from anonymous accounts, fabricated narratives spread faster than wildfires, and violence erupts on the streets, often ignited by the sparks of online vitriol. Platforms like Facebook, X, and WhatsApp, originally meant to connect, have become battlegrounds for bigotry, with their algorithms amplifying hateful narratives and turning echo chambers into cauldrons of rage.
This isn't some abstract concern. Look at the 2020 Delhi riots, fuelled by WhatsApp misinformation, where over 50 lives were lost in a religious inferno. See the 2021 attacks on Bangladeshi Hindus, sparked by a fabricated Facebook post, leaving temples desecrated and communities traumatised. Witness the 2019 Sri Lankan Easter bombings, where online hate speech may have played a role in the radicalisation of the attackers, or Facebook posts calling for violence against the Madhesi community sparking clashes and protests in Nepal. These are just a few grim examples of how the digital demon of social media is playing a deadly game in the form of fake digital nationalism in South Asia. What is the solution? It is important to identify and remove biased algorithms. The platforms have a responsibility to audit their algorithms for bias that amplifies hateful content and unfairly targets marginalised groups. Removing or correcting such biases can significantly reduce exposure to hate. It is also important to train AI systems to understand the context and intent behind language, distinguishing genuine criticism from hateful attacks. This requires ongoing investment in natural language processing and human oversight.
With so much hate in social media, there is a strong need to develop a spectrum of consequences for varying degrees of harm, ranging from warnings and temporary suspensions to permanent bans. There is also a need to encourage bystander intervention by making it simple and safe to report hateful behaviour. A 2020 study by the University of Oxford analysed over 250 million tweets related to India and Pakistan. The research found that 12 per cent of these tweets contained hate speech, with the majority targeting religious minorities and political opponents. #FreeRohingya has been used over 40 million times on X, highlighting the plight of the Rohingya Muslim minority in Myanmar. However, the online campaign has also been marred by calls for violence against Myanmar nationals and the spread of misinformation about the conflict. In the aftermath of the 2019 Easter bombings, social media platforms in Sri Lanka saw a surge in anti-Muslim hate speech. The hashtag #SinhalaRavaya (Sinhalese Power) gained traction, further dividing the country along ethnic lines. The same is true for the recent so-called student uprising in Bangladesh. Large number of fake videos were circulated contrary to what was actually happening. This trend should be curbed at all costs.
Views expressed are personal
