4/10/22, 10:31 Oversight Board | Independent Judgment. Transparency. Legitimacy. In response to questions from the Board, Meta explained that the content was only restored to the platform because it happened to be flagged by a Meta employee for an escalated level of review. “Escalated for review” means that, instead of the decision being revisited by at-scale review, which is often outsourced, it goes to an internal team at Meta. This appears to have required a Meta employee to notice the removal of the content, then fill out and submit an internal webform highlighting the issue. In addition to the element of chance, systems such as these can only identify errors in content to which Meta staff are personally exposed. Accordingly, content that is not in English, content not posted by accounts with many followers in the US, or content created for and by groups not well represented within Meta is far less likely to be noticed, flagged and given the additional attention. As part of its outreach, the Board was made aware of concerns from stakeholders that accurate enforcement of the exceptions to the Hate Speech policy requires a degree of subject-matter expertise and local knowledge that Meta may either lack or not always be able to apply. The Board shares concerns that, unless Meta regularly seeks input from minority groups targeted with slurs on a country-specific level, it will be unable to formulate nuanced lists of designated slur terms and give its moderators proper guidance on how exceptions to the slurs policy should be applied. 8.2 Compliance with Meta’s values The Board finds that the original decision to remove this content was inconsistent with Meta's values of "Voice" and "Dignity" and did not serve the value of "Safety." While it is consistent with Meta's values to prevent the use of slurs to abuse people on its platforms, the Board is concerned that Meta is not consistently applying exceptions in the policy to expression from marginalized groups. In the context of this case, “Voice” that seeks to promote free expression from members of a marginalized group is of the utmost importance. Meta is right to attempt to limit the use of slurs to denigrate and intimidate their targets, and also to allow good faith attempts to deprive those words of their negative impact through reclamation. The Board recognizes that the circulation of slurs impacts “Dignity.” Particularly when used with the intent to offend or absent contextual clues signifying that they are not being used to offend, encounters with slur words can intimidate, upset or offend users in ways that inhibit online expression. Where there are clear contextual clues that the slur is mentioned to condemn it, raise awareness for it, or mentioned self-referentially or in an empowering way, the value of "Dignity" does not dictate that the word must be removed from the platform. On the contrary, over-enforcement that ignores the exceptions particularly affects minority and marginalized groups. As recommended by the Board in the "Two buttons meme" decision, Meta must ensure that its moderators are sufficiently resourced and supported such that relevant context could be assessed properly. It is important that moderators are able to distinguish between permitted references to slurs and impermissible uses of slurs to protect the "Voice" and "Dignity" of its users, especially those from marginalized communities. As the “Dignity” and “Safety” of marginalized communities are at a heightened level of risk on social media platforms, those platforms have heightened responsibilities to protect them. The Board has already recommended in the "Wampum Belt" decision that Meta should conduct accuracy assessments on the application of Hate Speech policy allowances. Accuracy can be improved through the training of moderators so that they are able to identify content involving discriminated communities and receive instructions to carefully assess whether exceptions to the Hate Speech policy apply. An assessment of the content, along with supporting contextual cues, should be the triggering factor for the application of these exceptions. With regards to “Safety,” the Board also notes the particular importance of both safe online spaces and careful moderation to marginalized and threatened communities. LGBTQIA+ Arabic speakers, especially in the MENA region, face a degree of danger when openly expressing themselves online. Meta must balance the need to provide supportive arenas for this expression with ensuring that it does not over-moderate and silence people who already face censorship and oppression. While the Board acknowledges the complexity of moderation in this area, especially at scale, it is vital that platforms invest the resources required to do it properly. 8.3 Compliance with Meta’s human rights responsibilities The Board concludes that Meta’s initial decision to remove the content was inconsistent with its human rights responsibilities as a business. Meta has committed itself to respect human rights under the UN Guiding Principles on Business and Human Rights ( https://www.oversightboard.com/decision/IG-2PJ00L4T 7/11