面向攻击者的 LLM 越狱套件走向商业化,模型侧策略防护压力随之上升。
llm越狱工具
https://t.co/zKV2rbHvmH will continue to produce novel AI-driven hacking research. pro users get access to AI-kits and tutorials to help bypass content policy and keep doing their work. hackers never ask permission, neither will we.