Computer scientists at Northeastern University have tested a common complaint about TikTok — that its For You Page (FYP) ignores users' negative feedback — and found that users have less control than they believe. Their study concludes that engagement signals such as skipping videos or tapping "not interested" do work, but only temporarily: the algorithm gradually reverts to showing unwanted content unless users repeatedly reinforce the same feedback. The finding matters because the FYP is the platform's default screen, and it suggests that curating one's feed depends more on the platform's underlying design choices than on the tools offered to users.

The researchers, who specialise in "algorithm audits", created 90 emulated bot accounts on the real TikTok app, intercepting network traffic and using validated large language models to make viewing decisions, since official researcher APIs and EU data access only provide aggregated data unsuitable for studying personalisation. Testing three topics — cooking, fitness and sports betting — they found the "not interested" button cut unwanted content by around 84 per cent, compared with just 48 per cent for merely skipping videos, though they note the button appears deliberately hard to find. Even brief re-engagement caused the algorithm to "relapse" and flood the feed again. The work was published in the Proceedings of the Twentieth International AAAI Conference on Web and Social Media (2026).