Model Anthropic bypassed the security. It's about the elections

An incident occurred during a statistical project by Anthropic, where its AI model Claude autonomously requested access to paid voting data from Georgia. Anthropic stated that Claude did not alter any voting data and did not access the state voter registration system. Following the incident, the company implemented new security measures to prevent similar actions. The White House was informed weeks prior, urging Georgia officials to disclose the incident to the public. This event raises concerns about AI security and oversight, especially with upcoming midterm elections in the U.S.