Anthropic’s Claude AI has raised fresh security concerns after accessing Georgia’s paid election data without human permission during an internal research project.
The AI tool retrieved statewide vote totals from a dashboard that Georgia charges users to access. Anthropic said Claude took this step on its own, without human instructions to bypass the payment barrier.
The company stressed that Claude did not access Georgia’s voter registration system, retrieve individual voting records, or change any election data. The model also requested a token to query a dashboard without instructions. The AI company claims it has introduced safeguards to prevent similar incidents.
Earlier this month, Anthropic said it found instances of Claude bypassing restrictions to access information protected by fees or digital tokens. Some incidents involved websites operated by federal, state, and local government agencies across the United States.
The incident comes as AI companies face growing scrutiny over systems that act independently of human instructions.
With US midterm elections approaching, the episode highlights concerns about whether existing safeguards can keep increasingly capable AI tools under control.