Grok, the “anti-woke” chatbot from Elon Musk’s xAI, went on racist, violent tirades just a week before the company launched its own AI “companion” characters and won a $200 million American military contract. Is AI safety being taken seriously enough?