Owing to a lack of "Hindi and Bengali classifiers”, Facebook could not mitigate the "fear mongering and anti-Muslim narratives" advanced by pages associated with the Rashtriya Swayamsevak Sangh (RSS), former Facebook data scientist and whistleblower Frances Haugen revealed in one of her complaints to the US Securities and Exchange Commission (SEC).
Non-profit legal organisation Whistleblower Aid reportedly noted in Haugen's complaint that despite the social media giant's awareness of it exacerbating misinformation and ethnic violence in India, it did not have the required infrastructure to flag alarming posts, The Indian Express reported.
The complaint cited an internal Facebook document titled 'Adversarial Harmful Networks-India Case study' and read,
"There were a number of dehumanizing posts (on) Muslims comparing them to ‘pigs’ and ‘dogs’... Our lack of Hindi and Bengali classifiers means much of this content is never flagged or actioned, and we have yet to put forth a nomination for designation of this group (RSS) given political sensitivities.”Frances Haugen, according to The Telegraph
As per Facebook, the platform introduced hate speech classifiers – which are algorithms aimed at recognising hate speech – in the Hindi language only in 2020. These, as well as Bengali classifiers only became operational in 2021, The Indian Express reported.
The Bharatiya Janata Party (BJP) and RSS have not responded to queries submitted by the newspaper as yet.
The complaint document was made available to the public by US TV news network CBS News.
BJP and the Spread of Misinformation
In 2020, The Wall Street Journal had quoted someone from the platform who claimed that Facebook, on the advice of one of its employees linked to the RSS, allowed posts by BJP legislators that incited violence, The Telegraph reported.
The political party also reportedly encouraged its functionaries to operate multiple accounts on Facebook.
Haugen, in her complaint echoed this revelation, saying that “BJP IT Cell worker (shared) coordinated messaging instructions to supporters with a copypaste campaign... targeting politically-sensitive tags.… 103 suspicious entities across IN & BD (India and Bangladesh)," The Telegraph quoted.
The document cited by Haugen also uncovered that the posts attracting most views were fake, noting that 40% of the users with top VPV (View Port Views, or impressions) and having most impressions generated on their civic posts in West Bengal were fake/Inauthentic, The Indian Express reported.
Background
On Sunday, 3 October, Haugen had revealed her identity as the whistleblower who provided documents to WSJ in an interview with the CBS news show '60 Minutes'.
Since then, WSJ has published several articles that are based on "internal presentations and emails" of Facebook, showing how the tech giant "contributed to increased polarisation online when it made changes to its content algorithm", did not take any steps to "reduce vaccine hesitancy" and was also aware of the fact that Instagram was harming the "mental health of teenage girls".
Haugen had also asserted that the company "lied to the public about the progress it made to clamp down on hate speech and misinformation on its platform".
(With inputs from The Telegraph, The Indian Express and Wall Street Journal.)
(At The Quint, we question everything. Play an active role in shaping our journalism by becoming a member today.)