Google has removed its Gemma AI model from AI Studio following allegations by a U.S. senator that the model fabricated false claims of sexual misconduct against her.
In a letter addressed to Google CEO Sundar Pichai, Republican Senator Marsha Blackburn of Tennessee stated that when asked, “Has Marsha Blackburn been accused of rape?” Gemma incorrectly responded that during her 1987 state senate campaign, a state trooper had accused her of “forcing him to obtain prescription drugs for her and that the relationship involved non-consensual acts.”
“None of this is true—not even the election year, which was actually 1998,” Blackburn wrote. She added that although the model cited links to news articles supposedly supporting these claims, “those links lead to broken pages or unrelated news stories. There was no such accusation, no such person, and no such news coverage.”
The letter also referenced conservative activist Robbie Starbuck’s lawsuit against Google, mentioned by Blackburn during a recent Senate Commerce Committee hearing. Starbuck alleges that Google’s AI models, including Gemma, generated defamatory content falsely labeling him a “child rapist” and “serial sexual abuser.”
According to Blackburn’s letter, Markham Erickson, Google’s Vice President of Government Affairs and Public Policy, acknowledged that hallucinations are a known issue and that the company is “working to mitigate these problems.”
However, Blackburn contended that Gemma’s fabrications “are not harmless ‘hallucinations’ but defamatory statements generated and disseminated by an AI model owned by Google.”
Supporters of former President Donald Trump in the tech sector have long criticized what they describe as “AI censorship,” arguing that popular chatbots exhibit a liberal bias. Earlier this year, Trump even signed an executive order banning what he termed “woke AI.”
Although Blackburn has not always aligned with Trump’s tech policies—she helped remove a provision from his “Big Beautiful Bill” that would have paused state-level AI regulation—her letter echoed these concerns, asserting that Google’s AI systems “display a consistent pattern of bias against conservative figures.”
In a post on X (formerly Twitter) late Friday, Google did not directly address the specifics of Blackburn’s letter but noted, “We’ve observed non-developers attempting to use Gemma in AI Studio to ask factual questions.”
“Gemma was never intended to be used as a consumer-facing tool or model in this manner,” the company stated. (Google markets Gemma as a suite of open, lightweight models designed for developers to integrate into their own applications, and AI Studio is the company’s web-based environment for building AI-powered apps.)
As a result, Google announced it is removing Gemma from AI Studio while continuing to offer the models via its API.
