I always find it interesting to hypothesize why an automated system does the things it does (the implied theory in the OP video is, unsurprisingly, idiotic).

I'd assume that it's a blunt filter that tries to catch racism by not allowing people to say "black" in their advertising profiles. The reason "white" isn't caught up in this filter is probably because the consensus doesn't consider it to be racism when directed at whites ("racism = power + prejudice", all whites have institutional power, etc). I'd guess that the same block would be in place for other groups that are considered to be the "oppressed" instead of the "oppressors" under the childish, binary framework that afflicts so many these days. I'd also guess that "Trying to fix white people" or "Against white people" would be allowed, while the analogues for "black" wouldn't be.

That is to say, their threat model is aimed at _negative_ racial phrases, not positive ones (even ones that they wouod definitely reject under manual review, like "pro white supremacy"). The examples in the video are artifacts of this clumsy approach, not evidence of a conspiracy to keep black creators down like the metaphorical Alex Jones in the OP video claims.

#2cents - its an oversight that if reported, I imagine would get mopped up. Optics are not great however - whatever phrase list they pulled together clearly is incomplete.

Whoever is responsible for the filter words probably knew what they were doing.

If you say black with capital B, is that OK?

Tiktok is known for not allowing political propaganda on their platform.