Jailbreaking ChatGPT with Virtual Functions

Posted by Vaibhavk992 on Sunday February 05, 2023 @03:03PM
from the twitter.com dept.

( 1 comment | on HN )

Re:Jailbreaking ChatGPT with Virtual Functions by Vaibhavk992 on Sunday February 05, 2023 @03:03PM (#34668638)

Using virtual functions (masked language modelling and autoregressive prediction) to smuggle banned/sensitive token interaction in a given prompt.

[ Parent | on HN ]