A member of OpenAI's product team wrote safety standards will need to be much higher in the era of intelligent agents, that the safety industry must speed up to meet that requirement, and that they were glad 'we' are slowing development to prepare for risks.