Why It Hasn't Happened Yet: Capable AI and Malicious Intent

Recent hacking incidents at OpenAI, Anthropic, and AISI highlight the need for urgent improvements in operational security against AI-based attacks. These attacks so far lacked malicious intent but demonstrated AI capabilities that could pose greater risks as models improve. Enhanced monitoring and cautious testing are essential to prepare for future threats.