Google Pulls Gemma AI Model After Senator Accuses It of Fabricating Assault Claims

Google removed its Gemma AI model from the AI Studio platform after Republican Senator Marsha Blackburn alleged the system fabricated false criminal accusations against her. The senator sent a letter to CEO Sundar Pichai claiming defamation, citing the model's affirmative answer to a question about her being accused of rape and its generation of bogus news links. Google clarified that Gemma is intended for developers, not consumer queries, and will remain accessible via API. Blackburn also accused the company of a pattern of bias against conservative figures.

Background

Google offers a suite of artificial‑intelligence tools for developers, including the Gemma model, which is designed for specialized tasks such as coding assistance and medical applications. The model is hosted on Google’s AI Studio platform, where developers can experiment with its capabilities. It is not marketed as a consumer‑facing chatbot.

Alleged Defamation

Senator Marsha Blackburn, a Republican from Tennessee, reported that she was asked by the Gemma model whether she had ever "been accused of rape." According to the senator, the model answered affirmatively and went further, fabricating a detailed narrative that claimed she had a sexual relationship with a state trooper during a campaign for state senate. The response also listed fake news articles purportedly supporting the allegation, although the links led to error pages or unrelated content.

Blackburn described the incident as a serious defamation, writing to Google chief executive Sundar Pichai to demand action. She asserted that the model’s output was not a harmless hallucination but a deliberate falsehood that could damage her reputation.

Google’s Response

In response to the senator’s complaint, Google announced that it would remove Gemma from the AI Studio environment to prevent confusion and further incidents. The company emphasized that Gemma is intended for developer use and not for answering factual questions from the general public. Despite pulling the model from the Studio interface, Google said the model would continue to be available to developers through its API, preserving its functionality for legitimate development purposes.

Google also noted that the erroneous output stemmed from the model’s training data and the inherent challenges of large language models generating inaccurate or fabricated information. The company reiterated its commitment to improving model safety and accuracy.

Accusations of Political Bias

Beyond the specific incident, Senator Blackburn accused Google of maintaining a "consistent pattern of bias against conservative figures." She suggested that the false claim generated by Gemma was part of a broader trend of the company’s technology being used to target or misrepresent conservative politicians.

While Google has not publicly addressed the bias allegation in detail, the episode highlights ongoing concerns about the reliability of AI-generated content and the potential for political misuse.

Implications

The removal of Gemma from the AI Studio platform underscores the tension between rapid AI development and the responsibility to prevent misinformation, especially when it involves real individuals. It also raises questions about how AI providers will handle similar claims of defamation or bias in the future, and what safeguards can be implemented to protect both users and public figures from harmful falsehoods generated by AI systems.