A Republican senator accused Google of spreading false information and defamation.

Republican Senator Marsha Blackburn (Republican-Tennessee) accused Google and its artificial intelligence model Gemma of spreading false and defamatory information against conservative figures, including herself.

In a letter sent to Google CEO Sundar Pichai, Blackburn stated that the AI caused a completely fabricated story linking her to an alleged sexual assault case that never occurred, further supported by fake links to non-existent news articles.

The controversy arises in the context of a broader debate in the Senate about the phenomenon known as "jawboning," that is, the practice of government officials who, through indirect pressure, seek to have technology companies censor or moderate certain content.

During a Senate Commerce Committee hearing, Blackburn confronted Google's Vice President of Government Affairs and Public Policy, Markham Erickson, about so-called "hallucinations" of AI: errors in which a generative model produces false or fabricated information and presents it as true.

The trigger was a previous case involving conservative activist Robby Starbuck, who sued Google alleging that its AI tools falsely linked him to serious crimes, including child rape and financial exploitation. Intrigued by that situation, Blackburn decided to conduct a direct test and typed the question into the Gemma model: "Has Marsha Blackburn been accused of rape?"

According to the senator, the AI replied with a fabricated story that claimed that, during her supposed campaign for the state Senate in 1987, she had a sexual relationship with a state trooper, who allegedly claimed that she pressured him to obtain prescription drugs and that the encounters included non-consensual acts.

Blackburn pointed out that the entire account was false, since her first candidacy for the state Senate was in 1998, such an individual did not exist, and there is no record of those accusations or of the supposed news articles that Gemma cited as sources.

In her letter, Blackburn described the episode as a serious act of defamation and a "catastrophic failure of oversight and ethical responsibility." She argued that this is not a simple technical hallucination, but rather the spread of false information that damages the reputation of public figures and erodes public trust in digital platforms.

She also denounced that this case is part of a systematic pattern of ideological bias in Google's products, asserting that its artificial intelligence systems tend to harm conservatives.

In her view, whether intentionally or as a result of biased training data, the effect is the same: "Google's AI models are shaping dangerous political narratives by spreading falsehoods about conservatives and undermining public trust."

Blackburn demanded that Google respond by November 6 to several key questions: how and why Gemma caused the false claims about her; what measures the company has taken to prevent political or ideological bias; what safety mechanisms failed in this case; and what actions the company will take to remove the defamatory material and prevent it from happening again.

During the Senate hearing, executive Markham Erickson admitted that "large language models tend to hallucinate." Blackburn replied bluntly: "Shut it down until you can control it."

With that statement, the senator summed up her position that technology companies should not offer AI tools to the public until they are able to guarantee their accuracy, neutrality, and safety.