OpenAI's "Preparedness" team will hire AI researchers, computer scientists, national security experts and policy professionals to monitor the tech, continually test it and warn the company if it believes any of its AI capabilities are becoming dangerous.