TikTok accused of viralizing racist videos made with Google's AI

AI-generated racist videos are becoming a problem on TikTok . This is according to a study by Media Matters, a US media watchdog organization, which suggests that many of them may have been created using Google's new content generation tool, Veo 3 .

Among the videos located by the organization, content that associates black people with monkeys or criminals stands out; one of them has received more than 14.2 million views on the Chinese app.

"Users also post misleading AI-generated videos about immigrants and protesters, including videos of protesters being hit by cars," Media Matters notes. The association also highlights "particularly dystopian" content featuring "concentration camps and Ku Klux Klan attacks on Black Americans."

This type of content is prohibited by TikTok's policies. The social network prohibits the upload of videos that dehumanize racial and ethnic groups , as well as those that "threaten or express a desire to cause physical harm to a person or group." Regardless, this would not be the first time the site has experienced problems removing prohibited content. The company recently laid off part of its international team of moderators to delegate some of the work to artificial intelligence.

Indeed, as Media Matters points out, the videos appear to be mostly developed at least with Veo 3. The resulting results are very realistic, and the videos even have sound, a feature not available in the vast majority of AI-powered tools. Some of them also appear with Google's watermark in one corner.

As with TikTok, the US search engine prohibits the use of its technology to generate "harmful results." Typically, in these cases, Veo 3 would block the request outright; although it wouldn't be the first time a user has managed to "trick" the machine into producing content it shouldn't.

ABC has contacted both Google and TikTok for comment on Media Matters' investigation. Neither tech company has yet responded to the request.