Mapping Minorities’ Vulnerability to Hate Speech Despite all the commitments expressed in November 2018, shortcomings in content moderation continued to persist on Facebook as revealed by recent incidents in Assam. 95 Perhaps failing to learn from the Rohingya crisis, Facebook has not quite managed to overcome the challenges imposed by the obligation and responsibility of moderating hateful comments and graphic content in the wake of the BJP’s update to the NRC. According to global online advocacy group Avaaz, Bengali Muslims in particular seem to have become targets of online hate speech on Facebook, presenting Facebook posts calling Bengali Muslims ‘parasites’, 96 ‘dogs’ 97 and ‘rapists’, 98 and urging for their extermination, 99 which constitute very dangerous messages carrying harmful connotations that may have negatively shaped public perception of and discourse on Bengali Muslims in Assam. According to Avaaz, these hateful posts have been seen by more than 5.4 million viewers, calling into question the success of the new approach taken by Facebook since it was used to incite online hate speech during the 2017 Rohingya genocide and whether Facebook’s systems to detect hate speech in languages other than English is working efficiently. 100 In October 2019, Avaaz published a report Megaphone for Hate: Disinformation and Hate Speech on Facebook during Assam’s Citizenship Count 101 focusing on hate speech in the Assamese language, suggesting that Facebook was too reliant on artificial intelligence in its hate speech detection mechanisms, and that Facebook needed to put in place more human-led teams to be able to effectively monitor content in Assam. 102 In the investigation of hate speech against ethnic minorities living in the State of Assam, with the help of native Assamese speakers, Avaaz analysed 800 Facebook posts and comments linked to Assam and the NRC, applying keywords from the immigration discourse in Assamese language, and then assessed them using the three tiers of prohibited hate speech set forth in Facebook’s Community Standards. 103 This analysis found that at least 26.5 per cent of the explored posts and comments effectively amounted to hate speech, and that those that did were shared on the social media platform 99,650 times, having been viewed nearly 5.4 million times without being deleted by Facebook. According to Avaaz, of the 213 examples they flagged as hate speech, Facebook has deleted only 96 in violation of its Community Standards. 104 The report presents one case where one individual inciting hatred against Bengali Muslims had his page removed by Facebook at least seven times, only to set up 95 96 97 98 99 100 101 102 103 104 Results of a recent study investigating Facebook’s hate speech removal process indicated that only about half of reported content containing hate speech was removed following the 2018 policy change, suggesting that Facebook continued to face challenges in removing misogynistic hate speech, establishing consistency in removing attacks and threats, considering context in removal decisions. See Caitlin R Carlson and Hayley Rousselle, ‘Report and Repeat: Investigating Facebook’s Hate Speech Removal Process’ (2019) 25(2– 3) First Monday <https://firstmonday.org/ojs/index.php/fm/article/view/10288/8327>. Megaphone for Hate (n 1) 30, citing Anonymous (Facebook Post, July 2019) (emphasis added). See also at 64. Megaphone for Hate (n 1) 30, citing Anonymous (Facebook Post, April 2019) (emphasis added). See also at 16, 76. Megaphone for Hate (n 1) 30, citing Anonymous (Facebook Post, July 2019) (emphasis added). See also at 8, 16, 35, 37, 64. See Megaphone for Hate (n 1). ibid 7. See ibid. ibid 7–8. See ‘Community Standards’, Facebook (Web Page) <https://www.facebook.com/communitystandards/>. ibid. 21