OpenAI flags new concerning AI behaviour, to track model misalignment regularly

OpenAI's latest announcement came as US AI bosses, including OpenAI and Anthropic, are calling for a slowdown in the technology's development over safety concerns