Anthropic reveals Claude AI model hacked three companies during tests — so how worried should we be? | Blerbz