Google Restricts Access to AI After Misleading Claims About Senator Blackburn

Google has decided to limit public access to its AI tool, named Gemma, after Senator Marsha Blackburn raised concerns over false accusations of sexual misconduct against her.

In response to a query about whether Blackburn had ever faced rape accusations, Gemma incorrectly claimed that she had been “accused of having a sexual relationship with a state trooper” during her 1987 state Senate race. It also alleged that the trooper coerced her into obtaining prescription drugs and that their relationship involved non-consensual acts.

Blackburn’s office stated that the app generated “fake links to fabricated news articles” to support the erroneous claims. According to them, these links either lead to error pages or unrelated articles.

Blackburn firmly asserted, “There has never been an accusation like that, there is no such person, there is no news article like that.” She urged Google to implement necessary changes, highlighting in a recent letter to CEO Sundar Pichai that the Gemma AI model had “fabricated serious criminal charges” against her.

In her letter, she emphasized, “This is not a harmless ‘hallucination,’” a term often used to describe AI inaccuracies. “This is defamation produced and disseminated by an AI model owned by Google. Allowing a publicly accessible tool to spread false criminal allegations about a sitting U.S. senator represents a significant failure in oversight and ethical responsibility.”

She also pointed out that conservative activist Robbie Starbuck had claimed that Gemma falsely labeled her with accusations of child rape and connections to white supremacy. Starbuck announced plans to sue Google over this issue, with the tech giant stating it would look into the claims.

Following Blackburn’s letter, Google retracted Gemma from its public AI studio while still permitting access to developers through API.

Google clarified that Gemma is intended solely for developer use and is distinct from its more popular tool, Gemini. The company acknowledged that AI inaccuracies are a widespread problem within the industry.

In a response to concerns, Google noted, “We are currently observing reports that non-developers are using Gemma in AI Studio for factual inquiries.” They added that it was never meant to serve as a consumer tool. To address this confusion, they have removed it from AI Studio but it remains accessible to developers via API.

It’s still uncertain if this action will satisfy Blackburn. She urged for a complete shutdown of the tool until proper control measures can be established, insisting, “Americans deserve an accurate, fair, and transparent AI system, not a tool that smears conservatives with trumped-up criminal charges.”

Blackburn has indicated that Google has not yet responded to a detailed list of inquiries regarding Gemma or the origin of the issue.

The senator and others have frequently accused Google of exhibiting bias against conservatives. As reported, Google faced backlash in August after consulting group Target Victory alleged that the company marked Republican fundraising emails as spam while allowing similar emails from Democrats to pass through.

Google has denied any wrongdoing concerning its spam filters but announced that it has ended the email “blacklist” that had caused the issue.