Google has taken down its Gemma AI model from AI Studio after U.S. Senator Marsha Blackburn accused it of fabricating false claims about her.
In a letter addressed to Google CEO Sundar Pichai, the Tennessee Republican said the model falsely alleged that she had been accused of sexual misconduct during a 1987 state senate campaign, a story she called completely fabricated. Blackburn pointed out that Gemma’s response not only invented a nonexistent state trooper and baseless accusations but even got the year wrong, her campaign took place in 1998, not 1987.
According to Blackburn, the model also cited supposed news sources that led to broken links or unrelated pages. “There has never been such an accusation, there is no such individual, and there are no such news stories,” she wrote.
The senator’s letter referenced a similar case brought up during a recent Senate Commerce hearing involving conservative activist Robby Starbuck, who is suing Google after its AI models allegedly generated statements labeling him a “child rapist” and “serial sexual abuser.”
During that hearing, Google’s Vice President for Government Affairs and Public Policy, Markham Erickson, acknowledged the issue, explaining that such “hallucinations” are a known limitation of large language models. However, Blackburn countered that Gemma’s actions were far more serious, calling them “not a harmless hallucination” but “an act of defamation produced and distributed by a Google-owned AI model.”
Blackburn also criticized what she described as a broader pattern of bias against conservative figures across Google’s AI systems. Her comments align with growing concerns among Trump-aligned conservatives who argue that “AI censorship” often reflects political bias. Earlier this year, former President Donald Trump signed an executive order banning what he called “woke AI.”
Although Blackburn has not always supported Trump’s tech policies, she previously voted to remove a moratorium on state-level AI regulation from one of his proposed bills, she echoed those bias concerns in her letter to Google.
In response, Google posted a statement on X (formerly Twitter) on Friday night, without mentioning Blackburn directly. The company said it had “seen reports of non-developers trying to use Gemma in AI Studio and ask it factual questions,” emphasizing that the tool “was never intended for consumer use.”
Google described Gemma as a family of open, lightweight AI models designed for developers to integrate into their own applications. AI Studio, the platform where Gemma was being used, is meant to serve as a web-based workspace for building and testing AI-powered apps.
As a precaution, Google confirmed it is removing Gemma from AI Studio but will continue to make the models available through its API.
