TikTok faces scrutiny after an internal testing version of its new AI avatar tool became briefly accessible to users, allowing the creation of potentially harmful content. CNN’s Jon Sarlin discovered that the unintended release enabled users to generate videos featuring extremist rhetoric, including excerpts from Hitler’s “Mein Kampf” and Osama bin Laden’s “Letter to America.”

The AI tool, part of TikTok’s Symphony Digital Avatars suite, was designed to help businesses create product launch ads using AI-generated likenesses of paid actors. However, the version CNN accessed lacked moderation features and watermarks indicating AI-generated content, which are present in the officially released version.

TikTok spokesperson Laura Perez told The Verge that a “technical error” allowed “an extremely small number of users” to access this internal testing version for several days. The company has since removed access to the problematic tool.