I think certain populations have this effect, yes. As an example, teen suicides have genuinely risen in the US in the post-smartphone/post-social-media era. So I think the evidence (suicides are not subject to measurement error as much) is pretty strong that certain populations (all teenagers, for instance) are encountering unhappiness to a great degree.

But TikTok is renowned for having an algorithmically tailored feed that is specifically engagement maximizing. While there are some selection effects in the people one encounters in normal life, surely one must concede that an algorithmically tailored feed maximizing engagement cannot possibly be anything but highly selected.