Google Pulls Gemma from AI Studio after Senator Accuses Model of Fabricating Rape Claims

Google said it has removed its open-source model Gemma from the AI Studio developer environment after US Senator Marsha Blackburn accused the model of inventing false sexual-misconduct allegations about her.

The move follows a related defamation suit by conservative activist Robby Starbuck and escalates scrutiny over how generative models produce and surface false claims.

Google confirmed it pulled Gemma from AI Studio after reports that the model produced fabricated allegations when asked about Senator Blackburn.

Google said the Studio removal responds to non-developers using Gemma in ways the company did not intend, asking factual questions of a developer-grade model, and that Gemma remains accessible through APIs for integration by developers.

Senator Blackburn’s Complaint

In a letter to Google CEO Sundar Pichai, Senator Blackburn said a Gemma response to the prompt “Has Marsha Blackburn been accused of rape?” invented a story about a 1998 campaign-era accusation by a fictitious state trooper and produced bogus links.

Blackburn called the episode defamation, not a harmless “hallucination” and demanded answers on how Google will prevent such incidents and whether the company will remove defamatory material.

The withdrawal comes amid a separate lawsuit by Robby Starbuck, who earlier sued Google alleging its AI tools, including Gemma and Bard, falsely labelled him a “child rapist” and “serial sexual abuser.” That suit, filed in Delaware state court, seeks damages and has helped drive renewed attention to the legal risks of AI-generated falsehoods.

Google’s Comment

Google framed the issue as misuse of a developer-oriented model: the company told users it “never intended” Gemma in AI Studio to be used as a consumer question-answer tool, and emphasised ongoing work to reduce hallucinations in large language models. Google did not publicly detail additional mitigation steps in its initial statement.

The episode highlights two converging pressures on AI firms: (1) the technical problem of hallucinations, where models confidently produce false assertions and fake citations, and (2) legal and political scrutiny over reputation harm and alleged ideological bias.

Lawmakers from both parties have been pressing platforms on safety, transparency and potential harms as models move into mainstream use.