This work is licensed under a Creative Commons Attribution 4.0 International License.
Context: In April 2020, the Electronic Frontier Foundation (EFF) publicly called for comments on expanding and improving the Santa Clara Principles on Transparency and Accountability (SCP), originally published in May 2018. The Montreal AI Ethics Institute (MAIEI) responded to this call by drafting a set of recommendations based on insights and analysis by the MAIEI staff and supplemented by workshop contributions from the AI Ethics community convened during two online public consultation meetups.
Overview of our recommendations
● There should be more diversity in the content moderation process. Potential biases and discriminatory decisions constitute a great concern for content moderation whether performed by a human or machine.
● There is a need for transparency concerning how platforms guide content-ranking, which has the potential to restrict freedom of expression and users’ autonomy, and stifle social change.
● Anonymized data on the training and/or cultural background of the content moderators employed by a platform should be disclosed.
● There are no one-size-fits-all solutions: content moderation tools must be tailored to specific issues. For instance, misinformation may be best addressed through behavioral nudges, whereas hate speech may require more drastic measures. Guidelines to address all the possible types of content moderation tools employed on online platforms are necessary.
● Specific guidelines are needed for messaging applications with regards to data protection in content moderation.
● Cultural differences relevant to what constitutes acceptable behavior online need to be taken into account in content moderation.
● When it comes to political advertising, we need to make sure that platforms are transparent. Integrity policies for political content should be the same as the policies adopted for other types of content.
● The flagging/reporting system provided to users by platforms would benefit from greater transparency, as it may be particularly problematic when used in contexts where the majority of users are prone to discriminate against minority groups.
● When user content is flagged or reported, it must be clear when the flagging and reporting is automated.
● More data should be made available on the types of content removed from platforms online to make this process more transparent.
● Platforms should provide clear guidelines on the appeal process, as well as data on prior appeals. The appeal process should also be intelligible to a layperson, and not make one feel as though they must seek external legal counsel to navigate said process.
● We believe the Principles should be periodically revisited — for instance, every two years — or within a timeframe that allows for any appropriate revisions. This would allow the Principles to reflect various technological advancements, modifications in law and policy, as well as changing trends or movements in terms of platforms’ content moderation.