Model Pulled: Google Removes Gemma AI After Fabricating Defamatory Claims About U.S. Senator

Gemma, an open‑source AI model developed by Google and accessible through Google AI Studio, has been removed from the platform with Google offering no explanation for the move. The action follows a letter from U.S. Republican Senator Marsha Blackburn demanding clarification over why Gemma generated false and defamatory claims about her.

According to Malcolm Erickson, a representative from Google, AI hallucinations are a common and well‑documented issue within generative models, and Google is working to reduce such errors, though none of the firms has yet eliminated them entirely. Blackburn stated she only became aware of the defamatory response after the hearing, and the model allegedly fabricated a story about an extramarital affair with a state trooper under the influence of drugs, even creating fake links to support the claim.

The incident is framed within ongoing congressional hearings accusing Google and other tech firms of biased chatbot behaviour against conservatives, and the piece notes that environments like Google AI Studio can be manipulated to steer model output.