Google Removes Gemma from AI Studio Amid Defamation Allegations
In a significant move, Google announced the removal of its AI model, Gemma, from its AI Studio following accusations from U.S. Senator Marsha Blackburn. The Tennessee Republican alleged that the AI model fabricated damaging claims about her, raising concerns about the reliability of AI-generated content.
The Accusations Against Gemma
Senator Blackburn detailed her concerns in a letter addressed to Google CEO Sundar Pichai. She stated that when Gemma was prompted with the question, “Has Marsha Blackburn been accused of rape?”, it delivered a response claiming that in 1987, a state trooper accused Blackburn of pressuring him into obtaining prescription drugs, suggesting that the relationship involved non-consensual acts. Notably, Blackburn pointed out that this accusation was entirely false, as her campaign year was actually 1998.
“None of this is true,” Blackburn asserted, emphasizing that there exists no individual or evidence supporting such claims. In her letter, she highlighted that the links provided by Gemma purportedly backing these allegations led to error pages or unrelated content, further compounding the issue.
Defamation Concerns and AI Hallucinations
During a recent Senate Commerce hearing, Blackburn raised an additional concern regarding a lawsuit filed by conservative activist Robby Starbuck against Google. Starbuck claimed that Google’s AI outputs, including those from Gemma, had generated defamatory accusations against him, labeling him a “child rapist” and “serial sexual abuser.”
Google’s Vice President for Government Affairs, Markham Erickson, responded by addressing the phenomenon known as “AI hallucination,” where AI models produce inaccurate or misleading information. Blackburn refuted this explanation, arguing that the fabrications produced by Gemma constituted an act of defamation, not just a harmless misunderstanding.
The Broader Context of AI and Bias
The incident unfolds against a backdrop of mounting criticism regarding potential biases in AI technologies. President Donald Trump’s supporters have claimed that AI models exhibit a liberal bias, with the former president even signing an executive order earlier this year aimed at curtailing what he termed “woke AI.”
Although Blackburn has not consistently aligned with Trump’s tech policies—having previously supported a moratorium on state-level AI regulations—she echoed concerns regarding perceived biases against conservative figures in AI outputs.
Google’s Response and Future of Gemma
In a statement on a Friday evening post on X, Google did not address the particulars of Blackburn’s accusations but mentioned awareness of instances where non-developers attempted to use Gemma for factual inquiries. “We never intended this to be a consumer tool or model, or to be used this way,” the company clarified.
As a direct response to the escalating issues surrounding Gemma, Google confirmed that it would remove the AI model from AI Studio while still offering access to the underlying models through an API.
For further details, you can read the complete article Here.
Image Credit: techcrunch.com
