The latest GPT-5 model family, GPT-5.2, builds upon the safety mitigation approach of previous versions (GPT-5 and GPT-5.1). Like other OpenAI models, GPT-5.2 is trained on a wide range of data sources including publicly available information, data shared with partners, and user-generated content. This comprehensive approach ensures responsible development and use of advanced language technology.
מקור: OpenAI News — לכתבה המלאה
כתיבת תגובה