top of page

Why Facebook Needs to Update its Algorithm Regarding Language in Order to Prevent Hate Speech

There are currently 22 official languages of India. Taking Hindi and Bengali cumulatively, the figure comes close to breaching the mark of fifty per cent. However, a recent coverage by Indian Express has shown that Facebook that “classifiers for violent and incitement in Hindi and Bengali came online in early 2021.” This clearly means that there was hardly any algorithm to look into hate speech existing in Hindi and Bengali. It is equally possible that as of today there are no algorithms to detect hate speech in various other languages which are spoken widely in India.

It is indeed quite ironic that the company tends to not focus on ethical considerations related to India, in spite of the country being the largest market for Facebook in terms of the number of users. The sheer ignorance in ensuring basic standards of maintaining on the platform seems to have got ignored as their executives sit in a small confined place in the Silicon Valley or as Mr Bhaskar Chakravarti suggests in the piece that “Mark Zuckerberg and his crew have been quarantined in Silicon Valley from the very start and have created or acquired addictive social media products that captured the attention of billions in the global south from a safe distance.”

A recently released book titled ‘Constitutional Challenges in the Algorithmic Society’ offers clues on the importance of Algorithms and their impact on societies. As the book states “The fast-growing use of algorithms in fields of justice, policing, public welfare and the like could end in biased and erroneous decisions, boosting equality, discrimination, unfair consequences, and undermining constitutional rights such as privacy, freedom of expression and equality.” Taking this argument forward, a crucial component of disseminating information is the language that is used. However, if that language has already been corrupted with hate speech and violence, without any regulation or oversight, it loses its significance and leads to disorder. It is essential to keep updating algorithms as one of the studies carried out by Stanford University shows that AI struggles to recognise toxic speech on social media. As suggested by Federica Casarosa, in chapter 15 of the above-quoted book, human intervention can be required in ‘certain instances’ where AI fails to detect any hate speech content. At the same time, stifling of voices and branding it as a problem of ‘hate speech’ remains an ever-lasting issue.

Hate Speech Regulation in India

IT Rules,2021 provide ample scope for restricting hate speech in India. Under Section 3 of the Rules(Due Diligence by Intermediaries and Grievance Redressal Mechanism), a significant social media intermediary is required to ensure due diligence in cases such as where the information is patently false or untrue, threatens the unity and sovereignty of India, is harmful to the child and so on.


Holding Facebook accountable for lack of diligence should be seen as part of asserting ‘sovereignty of the nation’s laws and government over a digital company that possesses the ability to manipulate and change things at a large scale. Otherwise, it adds validity to the argument that the digital company is acting like what EIC did to the subcontinent, ruling according to its whims and fancies.

(Siddharth Chaturvedi is a student-scholar and researcher at Dharmashastra National Law University, India. He is also a research intern with Global Order).


bottom of page