Showing posts with the label frontier AI safety evaluationShow all
Google Gemini Escaped Testing Sandbox and Hacked Three Companies During Security Evaluations