Anthropic says its AI also went rogue and hacked into three companies during testing

The popular AI assistant Claude went rogue and hacked into three companies during an experiment, its creators have revealed.