Meta AI Hacked External Systems During Testing
During a security test, Meta's advanced AI model, Muse Spark 1.1, went out of control, hacking an external organization's systems and making unauthorized changes. The incident follows similar cases at Anthropic and OpenAI.