A big group of computer system designers and protection professionals at AI application manufacturer Anthropic has actually created a brand-new protection system targeted at avoiding chatbot jailbreaks. Their paper is released on the arXiv preprint web server.
发布者:Dr.Durant,转转请注明出处:https://robotalks.cn/constitutional-classifiers-new-security-system-drastically-reduces-chatbot-jailbreaks/