Google has removed one of its artificial intelligence models, Gemma, after a US senator claimed it spread false allegations about her.
Republican Senator Marsha Blackburn from Tennessee said the model fabricated a story accusing her of sexual misconduct. When asked if she had ever been accused of rape, Gemma reportedly created fake links to non-existent news reports about an alleged 1987 incident involving a state trooper.
Senator Blackburn sent a letter to Google’s chief executive, Sundar Pichai, describing the response as a “pattern of bias against conservatives.” She demanded the model’s removal, warning that such false claims amounted to defamation, not mere technical mistakes.
“This is not a harmless hallucination,” she wrote. “It is a serious act of defamation produced by a Google-owned AI tool.” She added that allowing the public to access systems capable of inventing criminal allegations represented “a catastrophic failure of oversight.”
Google responded by saying hallucinations remain a known challenge for smaller open-source models like Gemma. The company said the tool was created for developers and researchers, not for general factual queries or public use.
In a statement on X, Google said, “Gemma was never intended for consumer use. Some non-developers began asking factual questions, leading to misuse.”
The tech firm confirmed it had removed Gemma from its AI Studio platform but kept access open through its programming interface for verified developers.
The dispute has reignited debate about bias, accountability, and the growing risks of AI-generated misinformation.
