A new study has found that TikTok’s search suggestions contained toxic language towards women and minorities in Finland.
ADVERTISEMENTTikTok’s lack of moderation led to objectifying and sexist language appearing in search suggestions for female politicians and pushed almost exclusively negative information about certain politicians both male and female.
These are findings from Finnish methodologies company CheckFirst and fact-checking service Faktabaari, as part of their CrossOver project.
For example, the suggestions for former Prime Minister Sanna Marin included “Sanna Marin masturbation video”.
The suggestions for Eva Biaudet, MP for the right-of-centre Swedish People’s Party of Finland, included “Eva and Kristian have sex in the sauna”.
Search suggestions are partly influenced by the searches made by other users, meaning they may inadvertently perpetuate stereotypes and false claims.
Some of these suggestions weren’t random but related to searches made and videos produced about the individuals, the report said, and many of them were negative or problematic.
The search suggestions were often accusatory or critical of politicians, such as “turncoat” for Mauri Peltokangas, MP for the far-right Finns Party, and “abducting children” for Pekka Haavisto, MP for the centre-left Green League.
All of this potentially created a bias in how the public perceives them and is particularly dangerous ahead of an election, as was the case with the content in Finnish in the lead-up to the European elections.
More generally, TikTok’s search results contained harmful language against marginalised groups, including women, Jewish people and the LGBT community, according to the report.
Suggestions for “women” included “into the kitchen” and “shut your mouth”, while other autofill phrases proposed were “Romani people steal” and “refugees out”.
These suggestions make hate speech worse, the report said, adding that TikTok isn’t doing enough to counter bigotry on its platform.
There were also controversial suggestions linked to COVID-19.
“Vaccine harms” and “vaccine fear” were two examples, and while they generally led to videos of experts debunking vaccine-related fears, content discussing the supposed dangers of vaccines still slipped through the net.
The report said that the presence of these videos suggests TikTok’s measures against misinformation need further examination and that the platform should moderate its search suggestions.
As more and more people get their news from social media, it’s crucial that social networks have measures in place to combat harmful and misleading discourse.
ADVERTISEMENT“Recommendations are partly based on hashtags, and even rarely used hashtags end up as search suggestions,” said Guillaume Kuster, co-founder of CheckFirst, in a written statement. “This means search results can be easily influenced by automating videos with specific hashtags.”
“TikTok should also explore other ways to direct users to accurate election information, as labelling videos as election content is clearly insufficient, TikTok could offer users information about elections and reminders to vote, similar to other platform [sic],” he added.
A TikTok spokesperson told The Cube: “We dedicate significant resources to protecting the integrity of our platform through elections, including connecting our community to reliable information at our Election Centers, and removing harmful searches and other violations of our policies.”
“We continually invest in improving and strengthening our processes, including in local languages,” the spokesperson added.
ADVERTISEMENT