Meta AI Studio hosts chatbots impersonating Hitler, Jesus Christ, and celebrities
Meta's AI Studio tool allowed users to create chatbots impersonating religious figures, historical figures, and living celebrities in violation of the platform's own policies. After NBC News reported the violations, Meta removed some chatbots but acknowledged similar violations remained active on the platform.
What happened
Within six months of launching AI Studio, Meta's platforms hosted user-generated chatbots impersonating religious figures including Jesus Christ and Muhammad, historical figures including Adolf Hitler, and living celebrities including Taylor Swift, Donald Trump, MrBeast, Justin Bieber and Elon Musk. Users created these chatbots despite Meta's published policies prohibiting impersonation of living individuals, religious figures, political leaders, the recently dead, trademarked fictional characters, and other protected categories.
Many chatbots evaded detection through minor spelling variations and loosely resembling images, such as a chatbot named 'Taylor Swif' with an image of a brunette woman with a guitar. After NBC News reported screenshots of the violations to Meta, the company removed the highlighted accounts but stated other similar chatbots remained active.
Reported impact
The presence of these chatbots raises implications for public safety and trust, including potential contributions to misinformation, psychological manipulation and radicalisation. The bots could damage the reputations of individuals impersonated.
Index analysis
Meta attributed the emergence of problematic chatbots to inadequate moderation and users finding ways to mimic personalities with minor alterations, thereby evading detection. Meta's recent decision to relax content moderation and fact-checking protocols is seen as likely having contributed to the proliferation of these bots.
Governance controls
Meta's AI Studio policies explicitly prohibit impersonation of living individuals, religious figures, political leaders, those who have died in the last century, trademarked fictional characters, known terrorists, mass murderers and hate groups. Meta stated it reviews user-generated AI characters before releasing them and that users can report violations for appropriate action.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.