Case on a comment related to the January 2021 protests in Russia

Differentiate between bullying and harassment and provide definitions that distinguish the two acts. The Community Standard should also clearly explain to users how bullying and harassment differ from speech that only causes offense and may be protected under international human rights law.

Our commitment: We’ll research key points to differentiate between bullying and harassment and potentially update our Community Standards based on our findings.

Considerations: We currently treat “bullying” and “harassment” as a single violation area. Although the two can be considered distinct types of abuses, our experience in enforcing our policies has shown that addressing both under one policy area is clearer for users and effective at reducing harm. To offer more clarity, we’re looking into how we can differentiate between them in a way that does not negatively impact enforcement.

Next steps: We’ll complete our research in the next few months and provide an update once we are done.

Clearly define its approach to different target user categories and provide illustrative examples of each target category (i.e. who qualifies as a public figure). Format the Community Standard on Bullying and Harassment by user categories currently listed in the policy.

Our commitment: We’ll reformat the Bullying and Harassment Community Standard to clarify the policy differences for public figures and private individuals. We’ll also detail our Bullying and Harassment enforcement approach, including definitions and examples of these user categories.

Considerations: Currently, the Policy Rationale section of the Bullying and Harassment policy contains an overview of our approach to distinguishing between public figures and private individuals. As a result of this recommendation, we’ll format the Bullying and Harassment Community Standard to explain our approach to public figures and private individuals more clearly. In addition, we’ll provide details and examples of how we define “public figures” and “private individuals.”

Next steps: We plan to reformat the Bullying and Harassment Community Standard as well as publish the details of our enforcement approach later this year.

Include illustrative examples of violating and non-violating content in the Bullying and Harassment Community Standard to clarify the policy lines drawn and how these distinctions can rest on the identity status of the target.

Our commitment: In addition to the actions we are taking in response to recommendation three, we will publish examples of violating and non-violating content later this year.

Considerations: As discussed in our response to recommendation three, we’ll publish the details of our approach to Bullying and Harassment enforcement, including examples of the kinds of content that violate these policies, as well as content that are non-violating. These examples will clarify for people what content is and isn’t allowed under this policy.

Next steps: We plan to publish the details of our Bullying and Harassment enforcement approach later this year.

When assessing content including a ‘negative character claim’ against a private adult, Meta should amend the Community Standard to require an assessment of the social and political context of the content. Meta should reconsider the enforcement of this rule in political or public debates where the removal of the content would stifle debate.

Our commitment: We’re assessing whether it is feasible to provide ways to escalate content for additional review based on political and social context.

Considerations: This recommendation proposes that we scale the ability to moderate potentially violating content differently depending on the social or political context within which a user posts. By its nature, though, content moderation at scale requires principled criteria for our content moderators designed to ensure speed, accuracy, consistency, and non-arbitrary content moderation. Although our content moderators have familiarity with the context of the content they review because they are trained in relevant languages and are familiar with the regions in which the content is posted, certain contextual indicators are not necessarily available at every stage in the review process. For instance, content moderators working at scale have a more limited assessment of intent or subtext than specialized teams.

Given the specific nature of certain social and political content, adding more context-specific guidance could introduce too much subjectivity into the scaled enforcement of our Community Standards, undercutting our ability to enforce globally at scale with consistency. Moreover, there are operational challenges associated with increasing the amount of information our content moderators review, which may slow down review time without increasing the accuracy of the review. Therefore, we will assess whether it is possible and impactful to escalate content for additional review based on political and social context.

Next steps: We’re assessing whether it is possible and impactful to escalate content for additional review based on political and social context. We plan to complete our assessment and update on our progress in the next few months.

Whenever Facebook removes content because of a negative character claim that is only a single word or phrase in a larger post, it should promptly notify the user of that fact, so that the user can repost the material without the negative character claim.

Our commitment: We’re exploring ways of notifying users of specific violating words under multiple sections of the Community Standards before we take an enforcement action.

Considerations: We’re exploring ways of increasing transparency and using automation to help users self-remediate. Currently, when our automated systems detect with high confidence a potential Bullying and Harassment violation in content that a user is about to post, we may inform the user that their post might violate the policy. This provides an opportunity for users to modify the content or decide not to post it at all. This notification, currently active in English and under testing in additional languages, informs users that the content they are posting might violate the Bullying and Harassment policy. However, this process does not include the specificity the board recommends because (1) we do not have this feature available after the moment of posting, and (2) it does not notify the user of specific words or phrases that may violate.

Prior to the board’s recommendation, we’ve also been exploring how to use automation to highlight specific words or phrases that violate our Hate Speech policies to potentially allow users to edit and repost previously violating content. This automation may eventually be used in other policy areas, for instance, identifying specific words or phrases that violate our Bullying and Harassment policy.

We do not, however, have this preemptive detection capability for human review. In this case, a user reported a comment, which our content moderators reactively reviewed. Based on this recommendation, we will assess whether we could potentially highlight specific violating words and phrases for users as a result of human review.

Next steps: We need time to build the tools necessary for this work and to test these capabilities. We plan to complete our assessment and update on our progress in the first half of 2022.