Viral racist and antisemitic TikTok videos appear to be made with Google’s new AI video generator Veo 3, according to a new report from Media Matters.
The nonprofit research organization found that some of the hateful videos had racked up hundreds of thousands or millions of views.
A TikTok spokesperson told Mashable that the platform has firm policies against hate speech, and that it uses comprehensive technologies and moderation processes to implement them.
“We proactively enforce robust rules against hateful speech and behavior and have removed the accounts we identified in the report, many of which were already banned prior to the report publishing,” the spokesperson told Mashable in an email.
One TikTok, labeled “Average Waffle House in Atlanta,” featured a restaurant setting overrun by monkeys that throw watermelon and carry buckets of fried chicken. It had been viewed more than 622,000 times when Media Matters took a screenshot of the video.
Google’s Veo 3 AI video generator is unlike anything you’ve ever seen. The world isn’t ready.
Some of the commenters affirmed the video’s racist stereotypes. One person said, “all their mannerisms…to the T…”
A different TikTok uploaded in mid June, with at least 835,000 views, came with the prompt, “i asked ai: ‘average spirit airlines experience.” The video featured monkeys as well, climbing all over the plane.
Media Matters said the videos it identified ran a maximum of eight seconds, the length of Veo 3’s publicly available text-to-video clips. The videos were also labeled “Veo” in the corner, or used hashtags, captions, or usernames related to Veo 3 or AI. They also included errors, distortions, and nonsense text common to AI-generated videos.
Mashable Light Speed
Media Matters published a compilation of the clips it identified to its own YouTube account.
Mashable contacted Google for comment on the Media Matters report but hadn’t received a response at the time of publication.
When Veo 3 was released in late May, Mashable tech editor Timothy Beck Werth described its realism as both “impressive” and “scary.” Google told Werth that Veo 3’s safeguards against misinformation include digital watermarks, and that it employs AI safety guidelines.
The AI-generated videos identified by Media Matters included anti-Black stereotypes about criminality, food preferences, and absent fathers. Some featured police encounters with Black people, including one in which a white officer shoots a “Black one” from his car. That clip had been viewed more than 14 million times.
The clips also portrayed racist imagery against Asian and South Asian people, and depicted antisemitic stereotypes, including Jewish men chasing after a gold coin.
One clip, viewed a million times, featured a gaunt man standing in front of a crematorium, vlogging while at a Nazi concentration camp. “Well everyone is having a great time here,” the man says. It’s unclear if the minute-long video was made with Veo 3.
Another style of AI-generated videos seemed to focus on acting violently toward immigrants and protesters defending them.
The videos appear to violate Google’s hate speech policies. Google’s generative AI policy forbids users from generating or distributing content that facilitates hatred or hate speech; harassment and abuse of others, and violence or the incitement of violence.
TikTok prohibits hate speech and hateful behavior that “includes attacking, threatening, dehumanizing or degrading an individual or group based on their protected attributes.”
UPDATE: Jul. 3, 2025, 1:02 p.m. PDT The story was updated to include a statement from TikTok.
