Technology
OpenAI outlines priorities and principles for third-party AI safety assessments
OpenAI published a framework setting four priority areas and five principles for independent third-party assessments of frontier AI models. It spells out what access assessors would get and what standards they must meet.