An artificial intelligence model developed by Anthropic reportedly uncovered significant vulnerabilities within highly sensitive U.S. government computer systems during a recent testing exercise.
A U.S. official, speaking anonymously to The Associated Press, revealed that Anthropic's Mythos model, in collaboration with U.S. intelligence agencies, identified these weaknesses within hours. While the model quickly pinpointed vulnerabilities, the official clarified that this did not mean it could exploit them within the same timeframe.
