🖼️231231Prompt Injection as Role ConfusionHacker News·Prompt Injection as Role Confusion·3 months ago#l5hU3EhN#role-confusion#role#user#roles#reasoning#think+3 more🧰Tag tools✨Add tagLLMs can't tell who's speaking. We show they identify roles by writing style, not tags, and exploit this with CoT Forgery, injecting fake reasoning that models mistake for their own thoughts.15s0Read later0Read More
