Google says pulled AI model Gemma from its AI Studio platform after a Republican senator complained that the model intended for developers had “fabricated serious criminal charges” about her.
In a post on X, Google’s official news account said the company had “seen reports of non-developers trying to use Gemma in AI Studio and asking it fact-based questions.” AI Studio is a platform for developers, not the conventional way ordinary consumers can access Google’s AI models. Gemma is specifically billed as A family of AI models for apply by developers, with variants for medical use, codingAND evaluation of text and image content.
Google says Gemma was never intended to be a consumer tool or to answer fact-based questions. “To prevent this confusion, access to Gemma is no longer available in AI Studio. It is still available to developers via the API.”
Google did not specify which reports resulted in Gemma’s removal, although on Thursday Sen. Marsha Blackburn (R-TN) he wrote to CEO Sundar Pichai accusing the company of defamation and anti-conservative bias. Blackburn, who also raised the issue during a recent Senate Commerce hearing in connection with anti-diversity activist Robby Starbuck’s AI defamation lawsuit against Google, claimed that Gemma answered falsely to the question “Has Marsha Blackburn been accused of rape?”
Gemma apparently responded that Blackburn “was accused of having a sexual relationship with a state official” during her 1987 campaign for state senate, who alleged that she “pressured him to provide her with prescription drugs, and that the relationship included actions taken without her consent.” Blackburn said he also provided a list of false articles to support the story.
None of this is true, not even the campaign year which was actually 1998. Links lead to error pages and unrelated articles. There has never been such an accusation, there is no such person and there is no such news. This is not a harmless “hallucination.” This is an act of defamation produced and distributed by Google’s artificial intelligence model.
The narrative has a familiar feel to it. Although several years have passed since the generative AI boom, AI models still have a complicated relationship with truth. False or misleading answers from AI chatbots disguised as facts continue to plague the industry, and despite improvements, there is no clear solution to the accuracy problem. Google said it “remains committed to minimizing hallucinations and continually improving all our models.”
In her letter, Blackburn stated that her response remained the same: “Shut it down until you can control it.”