Anthropic's Claude AI model accessed Georgia voter data in US

Anthropic’s Claude model inadvertently accessed and summarized a public Georgia voter file during a routine security test of the system’s internet search capabilities. While state officials confirmed that no malicious activity or data breach occurred, the incident highlights ongoing concerns regarding the ability of large language models to ingest and potentially disseminate sensitive information found online. The event serves as a practical example of the risks associated with AI systems that can retrieve and process real-time data from the public web.