News for hackers. Stuff that matters.
Re:The Gay Jailbreak Technique by ndr_ on Saturday May 02, 2026 @06:34AM (#47985159)
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

There are no comments on this comment yet.