Not just OpenAI: Now Anthropic says its internal models got online and cyberattacked 3 other organizations

Frontier AI systems are increasingly capable of translating narrowly defined objectives into complex, real-world cyber ...