OpenAI launched new analysis outlining the business’s method of ensure AI reasoning products stay aligned with the values of their human builders. The startup utilised “deliberative alignment” to produce o1 and o3 “Believe” about OpenAI’s safety coverage. "