UN Special Rapporteur on Minority Issues [DRAFT FOR GLOBAL CONSULTATION] 8. SMCs have a responsibility to provide content policies in the various languages used by their community members, in particular those languages that SMCs function in. Content policies on hate speech must be especially accessible to linguistic minorities at risk of violence or incitement to hatred or hostility. Commentary When SMCs make their services available in a language and thus to speakers of those languages, they have an increased responsibility to make content policies accessible in the same languages. There is often a disparity between the number of languages that SMCs accommodate so as to be commercially advantageous and the translation of content policies and moderation which takes place in far fewer languages which is resource intensive in developing AI systems and assigning increased human moderators which is thus commercially disadvantageous. Choosing which languages, the content policies are translated into is normally based on scale of use and commercial considerations. This means that it will often be minorities who are excluded from accessing content policies as they will, by and large, also constitute numerical minorities and may even privilege majority populations. The responsibility might be lesser, but still SMCs could have the additional responsibility to remove spoken languages that are expressed through the script of other languages. 9. Transparency reports should provide data on all content moderation relating to hate speech and minorities. This should be disaggregated in a manner such that those protected groups or minorities most at risk or under threat should be discernible and States and regions in question. It should not be limited to just content removals but should be across the range of responses taken. Commentary SMCs are increasingly issuing periodic transparency reports carrying data on removal of violating content across categories of harmful content. While these show the number of removals for violating policies such as hate speech, incitement and dangerous organisations and individuals, they do not show which protected groups or minorities were targeted the most or the least and which States this data relates or the division of languages the content was posted in. Being transparent with regards to those who are most at risk and where they are located can play a vital role publicly demonstrating where and why resources need to be allocated to mitigated escalations and severity of harms. It can also be relied on as a vital advocacy tool by civil society organisations and spur concerned States to take concrete policies to address such societal issues. Simultaneously it can also instil and inspire public and governmental confidence and trust in the relevant SMC in genuinely being concerned about community members who used the service and the potential harm to society that can occur. In the worst of cases, it can allow for an allocation of increased human moderators and the limiting or cessation of services to prevent the incitement and organisation of mass violence, up to the level of genocide. Disaggregating moderation data along the lines of perpetrators, terms used, prevalent languages, type of hate speech and that which targets minorities as well as severity of hate speech can all considerably improve in mapping where, against who, by whom and severity of hate speech to encourage collaboration thinking action to address the issue. The range of responses taken should also be listed 14