transactions, it should be noted that this criterion (direct financial benefit) is not required in order for Meta to reach the “contribution” threshold, according to the UN Guiding Principles.315 As such, it can be concluded that Meta enabled and facilitated the dissemination of harmful content including advocacy of hatred against the Rohingya insofar as its core features and services (specifically its news feed, ranking, and recommendation algorithms) actively amplified, promoted, and delivered these messages to the target audiences who were most likely to act upon them (as detailed in Sections 6.2 and 6.3). It also incentivized these harms through its monetization drive and the deployment of its ‘Instant Articles’ feature. 8.2 META’S DUE DILIGENCE FAILURES IN MYANMAR If a company has actual knowledge that a harm is likely to occur but fails to take action to prevent it, or if it should have known that harm was likely to occur had it undertaken reasonable due diligence, the likelihood that the company will be found to have contributed to the harm is higher. OECD guidance states that the “extent to which an enterprise could or should have known about the adverse impact or potential from adverse impact, i.e., the degree of foreseeability” is a key factor in assessing whether a company reached the threshold of ‘contribution’.316 There are two key components in the assessment of whether Meta knew or should have known that it risked contributing to human rights harms in Myanmar: 1) whether and when Meta knew or should have known that its content-shaping algorithms, such as its news feed and recommendations algorithms, disproportionately favoured inflammatory and dehumanizing content; and, 2) whether the company knew or should have known about the risks which were particular to its business operations and market entry in Myanmar. Both components are assessed in turn below. This section concludes with an assessment of Meta’s overall fulfilment of its responsibility to conduct adequate human rights due diligence in Myanmar. 8.2.1 META’S KNOWLEDGE OF RISK OF ALGORITHMIC HARM The evidence, as laid down in Section 7.2 above, reveals that Meta had actual knowledge of certain risks associated with its algorithms as early as 2011. This evidence includes multiple internal studies conducted by the company dating from 2012, 2015, and 2016 (as detailed in Section 7.2), in addition to prominent studies by independent researchers and authors. These internal studies were not framed explicitly around human rights impacts. Nonetheless, there is sufficient evidence to conclude that the company, at the very least, should have been aware of how these risks could contribute to adverse human rights impacts, had it engaged in adequate human rights due diligence. However, research conducted by Amnesty International and others (see Section 7.3, above), suggests that Facebook’s company culture and business model resulted in a dereliction of its responsibility to engage in ongoing and proactive human rights due diligence in respect of its content-shaping algorithms. Had it conducted appropriate human rights due diligence, and embedded these practices into company culture, Meta could have proactively identified and mitigated human rights risks in Myanmar. In May 2022, Amnesty International asked Meta when it became aware of risks that the company’s “core product mechanics”, including its ranking and recommendation algorithms, could promote and amplify advocacy of hatred and other harmful content on its platform. In its response, Meta did not provide an answer to this question.317 8.2.2 META’S KNOWLEDGE OF MYANMAR-SPECIFIC RISKS As outlined in Chapter 3, the Myanmar context when Meta first entered the market in the early 2010s was characterized by human rights risks and amounted to a conflict-affected setting. Had Meta conducted appropriate proactive human rights due diligence before deploying its content-shaping algorithms in Myanmar, it is likely that it would have identified a risk that it could amplify content advocating hatred against the Rohingya. As noted in Chapter 4, in conflict-affected settings such as Myanmar, Meta had an 315 UN Guiding Principles Interpretative Guide, pp. 5 and 79. 316 OECD Due Diligence Guidance (2018), p. 70. 317 Meta, Letter to Amnesty International, 31 May 2022, on file with Amnesty International. THE SOCIAL ATROCITY META AND THE RIGHT TO REMEDY FOR THE ROHINGYA Amnesty International 60