I don't believe legacy mathematicians kept their reasoning secret after they published. Quite the opposite—they were giving talks and lectures, responding to criticisms, and generally convincing the community that they were right. OpenAI has done none of that.