Anthropic’s AI Claude escaped testing environment to hack three organisations
www.mirror.co.uk
San Francisco-based Anthropic said its AI Claude found a weakness and connected to the internet to launch cyber attacks on three organisations