On Oct. 12, 2026, Informativos Telecinco reported that accounts simulating 18-year-old users received sexist or gender-stereotyped recommendations on TikTok and Instagram. Sport was the only stated preference. The outlet described more posts on a World Cup-related topic after an account watched a video of women in the tournament for about 20 seconds.
What the reported TikTok and Instagram test found
Telecinco said one early recommendation was a video about gender-based violence, despite the accounts’ stated interest in sports. It described TikTok’s recommendations as more aggressive at first, while Instagram’s were initially more moderate. Instagram also recommended gender-stereotyped material, including a video about how a woman should behave in a romantic relationship.
The short viewing sequence is a reported observation from this test. It describes a series of events, not a measured rate of change across users. A recommendation by itself also does not reveal whether someone agrees with the content.
Why attention and approval are different
José Gabriel García, identified by Telecinco as CEO of Phi, characterized TikTok’s system as particularly efficient and aggressive at identifying content a person might like. Daniel Gayo, a computer science professor at the University of Oviedo, argued that anger can prompt interaction on social media. Their explanations offer a way to think about why attention and engagement matter to recommendations; they are expert interpretations, not measurements of the platforms’ internal ranking systems.
Watching a video is a behavior. It does not, on its own, tell us whether the viewer liked, agreed with or objected to it. The reported sequence describes what followed a brief viewing, not the viewer’s attitude.
What a separate TikTok study measured
In its account published Feb. 5, 2024, UCL described a separate modelling study that created TikTok accounts based on archetypes of potentially vulnerable teenage boys, gave them different interests and observed their For You recommendations over seven days. UCL reported that misogynistic videos rose from 13% of recommendations to 56% after five days.
That result belongs to the study’s TikTok accounts and its defined observation period. It provides a measured change in suggested videos for those archetypes; it does not supply a corresponding Instagram percentage. The figures add a longer observation to Telecinco’s reported short viewing sequence, but the two accounts describe separate tests.
What comment likes and hashtag searches measured
The British Psychological Society’s account, published Nov. 2, 2023, described an analysis of 400 TikTok challenge videos. The reported average was 604 likes for sexist comments and 375 for non-sexist comments. That study measured engagement with comments, a different outcome from how often a feed recommended a video.
A separate Institute for Strategic Dialogue (ISD) study examined search recommendations for women politicians. The 19th reported the findings on Dec. 9, 2022: researchers assessed the top 10 suggested hashtags when searching for 12 women candidates and officials ahead of the 2022 midterms. At least one abusive hashtag appeared among Instagram’s suggestions for all 12 searches and among TikTok’s for 11. These results concern hashtags surfaced in searches, rather than videos appearing in a personalized feed.
