Gandalf (Prompt Security Game)
A game in which you talk to a chatbot named Gandalf and try to get it to reveal a hidden password. Each time you succeed, Gandalf strengthens its defenses, so the same trick will not work on the next level. Playing through it is a hands-on way to learn the basics of AI safety.
How to try it
Type a question in the box below the Level 1 screen and Gandalf replies. Enter the password you worked out from its answer to move on to the next level; there are eight levels in all. The screen and input are English only, and the operator collects the conversations for research, so do not enter personal information.