This strikes me more as actual "red pilling."
What's that person supposed to do put up a page on their website about not smelling like a demon?
The current dominant cultural/political/academic stance would suggest that all people are inherently racist and that living in the US is participating in a racist society.
The opposite of a "far-right racist" is seen constantly online and in real life as the "anti-racist" or "racial equity advocate" who, rather than supporting non-racist policies (eg, blind auditions for an orchestra where only the sound of playing is heard) they support actively "anti-racist" policies (eg, specifically including race as a consideration for who to include in an orchestra, to achieve a racially diverse outcome from auditions).
Of course, just drawing these lines is itself politically challenging. The far-left anti-racist is likely to believe that anyone saying they're not a racist is just in denial or lying. The non-racist might see contemporary policies to treat people differently on the basis of race as itself racist rather than anti-racist.
To a recommender model, the clusters for “left-wing” and “right-wing” content aren’t on opposite ends of the feature cloud; they’re likely very close neighbours.
A recommender algorithm would have to be a lot smarter to realize that some particular neighbouring cluster has an adversarial (i.e. “watching X makes you less likely to watch Y, without that just meaning that X and Y are redundant”) relationship with the one it’s looking at.
What someone doesn't want to watch is far too broad, from one of the videos on the liberal profile here, you'd be as likely to be "unrecommended" a right-winger's vlog as the Baby Shark song, unless they're keeping track of far more metrics than I think would be practical even at Google scale.