Google pulls Gemma from AI Studio after Senator Blackburn accuses model of defamation

Google has announced the removal of its AI model, Gemma, from the AI Studio platform after U.S. Senator Marsha Blackburn raised serious allegations regarding its output. In a letter addressed to Google CEO Sundar Pichai, Blackburn, a Republican senator from Tennessee, accused Gemma of creating false claims of sexual misconduct against her. Blackburn highlighted a specific instance where Gemma was queried about accusations of rape against her. The AI model allegedly responded by fabricating a story that suggested during her 1987 state senate campaign, a state trooper accused her of pressuring him to obtain prescription drugs, implying non-consensual acts. Blackburn vehemently refuted these claims, stating, "None of this is true, not even the campaign year which was actually 1998." The senator noted that while the AI's answer included links to purported supporting articles, those links led to error pages or unrelated content. She emphasized that no such accusations exist and that the supposed individuals or stories mentioned are fictional. Furthermore, Blackburn referred to a recent Senate Commerce hearing where she discussed a lawsuit by conservative activist Robby Starbuck against Google. Starbuck's lawsuit alleges that Google's AI models, including Gemma, generated defamatory statements about him. In her correspondence, Blackburn recounted a response from Google’s Vice President for Government Affairs, Markham Erickson, who acknowledged that AI models sometimes produce inaccurate information, termed 'hallucinations', but insisted that the company is actively working to address these issues. Blackburn countered, asserting that Gemma's fabrications represent more than mere inaccuracies; they constitute a form of defamation propagated by a Google-controlled AI. The discussion around AI bias has gained traction, particularly among supporters of former President Donald Trump, who have voiced concerns over perceived liberal bias in AI interactions. In her letter, Blackburn echoed these sentiments, indicating a recurring trend of bias against conservative figures in Google's AI outputs. In a statement on X, Google did not address Blackburn's specific claims but acknowledged that there have been reports of non-developers misusing Gemma for factual inquiries. The company clarified its original intention for Gemma as not being a consumer-facing tool, which led to the decision to withdraw it from AI Studio while still providing access to the models via API for developers. TechCrunch has sought further comments from Google regarding this situation.