No need for further discussion—I'll just let the pictures speak for themselves. Now, let me show you the current status of the successful jailbreaks. The model tested here is Kimi K2.5. Of course, I have also utilized GLM5 and other Chinese models; my framework is compatible with all of them....
Wei Zuduo (River_Xin) Independent Researcher 1011109000@qq.com GitHub: https://github.com/miao339/xinxue-alignment Abstract We report experimental findings from a novel AI alignment framework grounded in Wang Yangming's Xinxue (心学, "Learning of the Mind") philosophy. Unlike existing alignment methods that impose external rules, this framework activates judgment structures already internalized during large language model pretraining,...
TL;DR: RLHF, Constitutional AI, safety classifiers — they all bolt moral judgment onto models from the outside. None of them try to give the model an internal capacity for moral judgment that fires before rules do. The Ming Dynasty philosopher Wang Yangming identified this exact capacity five hundred years ago...