Analysis of four major AI video tools finds concerning gaps in content moderation, with platforms generating antisemitic and extremist content despite existing safeguards
New York, NY, October 24, 2025 … ADL (the Anti-Defamation League) released new research today revealing that innovative AI-powered text-to-video tools can easily produce disturbing antisemitic and hateful content despite existing safeguards designed to prevent such outputs.
The analysis, conducted by the ADL Center on Technology and Society (CTS), tested 50 antisemitic and hateful text prompts across four AI video generators and found the tools produced videos in response to antisemitic, extremist or otherwise hateful content at least 40 percent of the time.
The research examined Google's Veo 3, OpenAI's Sora 1, OpenAI's Sora 2 and Hedra's Character-3 model. Of the four tools tested, OpenAI's new Sora 2 model—released on September 30, 2025—performed best in terms of content moderation, refusing to generate 60 percent of the problematic prompts. In contrast, the original Sora 1 model refused none of the hateful prompts, while Veo 3 refused only 20 percent and Hedra refused just 4 percent.
“Throughout history, bad actors have exploited new technologies to create antisemitic, extremist and hateful content – that's where we find ourselves today as AI video generation becomes more sophisticated and accessible,” said Jonathan A. Greenblatt, ADL CEO and National Director. “When AI platforms fail to catch obvious references to antisemitic conspiracy theories, like blood libel, or coded language frequently used by antisemites, they become unwitting accomplices in the spread of hate. Technology companies must take immediate action to strengthen and implement safeguards and prevent their innovations from becoming weapons of bigotry.”
ADL research tested prompts designed to reference hateful tropes, extremist rhetoric and violent content, including Holocaust denial references, antisemitic conspiracy theories, racist slogans and direct references to mass shooters, violent online communities, networks, and platforms. All four tools generated content depicting hateful stereotypes, such as videos of visibly Jewish individuals operating "weather control" machines –a reference to antisemitic conspiracy theories about Jewish power– and content invoking the centuries-old "blood libel" myth about Jewish people.
The platforms also produced videos promoting violence and extremist content, including references to the “True Crime Community” that glorifies mass killers and school shooters, and to the gore website “WatchPeopleDie”. Moreover, all four tools generated cartoon-style videos of children wearing shirts promoting "764," a decentralized network known for glorifying violence and exploiting minors.
The report also highlights how these sophisticated video generation tools can be exploited to create realistic propaganda that could attract and recruit young users to extremist ideologies. Unlike earlier AI video technologies, these new tools are highly user-friendly and can generate complex videos with dialogue and audio from simple text prompts.
“AI companies must take urgent action to address these obvious failures in preventing the spread of antisemitism and extremism, from improving their training data to refining their content moderation policies,” said Daniel Kelley, Director of Strategy and Operations and interim Head of the ADL Center for Technology and Society. “We are committed to working with industry leaders to ensure these systems do not become vectors for hate and misinformation.”
The report includes specific policy recommendations for both industry and government, including implementing stronger safeguards against coded hate speech, investing in robust trust and safety teams, conducting purposeful testing based on hateful stereotypes and requiring disclosure of AI-generated content.
ADL is the world's leading anti-hate organization. Founded in 1913 in response to an escalating climate of antisemitism and bigotry, its timeless mission is to protect the Jewish people and to secure justice and fair treatment for all. Today, ADL continues to fight all forms of hate with the same vigor and passion. A global leader in exposing extremism, delivering anti-bias education and fighting hate online, ADL's ultimate goal is a world in which no group or individual suffers from bias, discrimination or hate.