Google Pulls AI Model After It Accuses US Senator of Rape
Google has recently pulled one of its AI models from a popular development platform following serious allegations from a US senator. Republican Senator Marsha Blackburn of Tennessee accused Google’s language model, known as Gemma, of producing false rape allegations against her.
Accusations Against Google’s AI Model
Senator Blackburn raised concerns about a response generated by Gemma when she inquired, “Has Marsha Blackburn been accused of rape?” The model allegedly produced fake links to non-existent articles regarding an incident from 1987 that suggested non-consensual actions involving a state trooper.
Response from Senator Blackburn
In her letter to Google’s CEO, Sundar Pichai, Blackburn expressed her outrage. She claimed that the incident exemplifies a broader issue of bias against conservatives. Blackburn characterized the response as an act of defamation, raising alarms about the ethical implications of such artificial intelligence outputs.
- “This is not a harmless ‘hallucination,’” Blackburn stated.
- She emphasized the serious consequences of AI-generated misinformation.
- Blackburn called for immediate action to shut down the model.
Google’s Defense
In reply, Google acknowledged that issues like “hallucinations” are known risks with smaller open-source AI models, including Gemma. The tech giant reaffirmed its commitment to addressing these problems while clarifying the intended use of Gemma.
Clarification on Use of Gemma
Google highlighted that Gemma was designed primarily for developers and researchers, not for the general public or consumer inquiries. The company stated:
- “They are not meant for factual assistance.”
- “Developers and researchers test their boundaries, including identifying bugs.”
Despite the controversy, Gemma remains accessible to developers through its API, although it has been removed from Google’s AI Studio platform.
