Artificial intelligence models that rely on human feedback to ensure that their outputs are harmless and helpful may be universally vulnerable to so-called ‘poison’ attacks.
![](https://wecoachcrypto.com/wp-content/uploads/2023/11/0f6e62fe-da13-4a0c-af77-fa1db8195a7a-IxZTDH.jpeg)
Artificial intelligence models that rely on human feedback to ensure that their outputs are harmless and helpful may be universally vulnerable to so-called ‘poison’ attacks.