DCU Anti-Bullying Centre — Recommending toxicity: algorithmic recommender systems

What the study found

The April 2024 report Recommending Toxicity, produced by researchers at the DCU Anti-Bullying Centre and funded by Ireland's Department of Justice, tracked the content recommended to 10 experimental "sockpuppet" accounts set up on blank smartphones — five on TikTok and five on YouTube Shorts, registered as 16- and 18-year-old boys. Every male-identified account was fed masculinist, anti-feminist or other extremist content within the first 23 minutes, regardless of whether the account sought out such material, and by the final round of the experiment the recommended content was on average 76 per cent toxic on TikTok and 78 per cent on YouTube Shorts. ^[1]

Why it matters here

This study is often cited as evidence that exposure to manosphere and male-supremacist content can be driven by platform recommender systems rather than only by what a user actively searches for. Because the accounts were newly created and, in several cases, did not seek out such material, the findings speak to how algorithmic amplification — not just individual choice — may shape what young male users encounter online. That mechanism is central to the research of Debbie Ging, one of the report's authors and a leading academic voice on the manosphere.