OpenAI trained o1 and o3 for safety policy

2 points| nancy_le | 1 year ago |techcrunch.com