Yes. OpenAI's GPT-OSS was training using Deliberative Alignment (which was found to be flawed in a competition on Kaggle, but still). https://arxiv.org/abs/241

Yes. OpenAI's GPT-OSS was training using Deliberative Alignment (which was found to be flawed in a competition on Kaggle, but still). https://arxiv.org/abs/2412.16339 https://arxiv.org/abs/2412.16339