Our team has been conducting research regarding the prevalence of eating disorder related content on social media. In particular, we have examined the ways users of TikTok discover eating disorder content, especially pro-eating disorder content.
One of the primary ways TikTok helps to curb the discovery of pro-ed content is through the use of a block-list filter, which filters keywords used in the various places users may search within the app. Our research indicates that there are keywords that are missing from this filter, many of which have substantial associated video counts.
We believe that by crowd-sourcing keyword candidates for inclusion in social media block lists, we might provide a helpful resource to social media platforms like TikTok. As platxaforms update their block lists, users make attempts to subvert those updates by modifying keywords, using homoglyphs or character substitutions, or in some cases associating a random or tangentially related word as a surrogate for a blocked keyword.
It is the community of users who are most aware of how filter evasion is happening, and therefore they are the most likely source for maintaining a robust block list. It is our hope that this experiment, which combines AI evaluation of keywords with human review, will allow us to present keyword block-list candidates to social media companies like TikTok, providing them with a powerful tool in the efforts to make their platforms safe for their users.
As a company with a deep understanding of the impact and dangers related to exposure to pro-eating disorder related content, we're excited to begin this experiment and assess the viability of such an approach.
If you have a press inquiry regarding this article, please use our Press Contact Form.