JUST IN: OpenAI shuts down team responsible for assessing whether its AI models could pose "catastrophic risks," FT reports.