Skip to Content Skip to Content

A team of roboticists at Penn showed that “jailbreaks” can cause AI models to misbehave, with unexpected and serious consequences when the model operates a robot.

https://tinyurl.com/38h3mnm3 Wired