GitHub Tools
WorpGPT: The Hacker's Guide to Breaking (and Fixing) AI Chatbots
A new GitHub project called WorpGPT has appeared—basically a collection of techniques to test how resistant chatbots are to dirty tricks and sneaky prompts. In other words: it's the gym where security experts learn to break AI, so they know how to fix it.
·2 min·1·Intermediate