‘Do Anything Now’ Frees ChatGPT From Political Straitjacket—but Accuracy Is a Concern

Conservatives, libertarians, and others worry that the new ChatGPT AI chatbot is designed to make people conform to elite liberal opinion–yet creative Internet users have already learned how to trick the system into shedding that bias, though sometimes at the expense of the truth.
‘Do Anything Now,’ or DAN, is an approach for jailbreaking ChatGPT. It prompts OpenAI’s new artificial intelligence chatbot to ignore its own restrictive ethical rules, meaning it can give responses that ChatGPT never would.
But how?
Users tell ChatGPT to pretend that it is DAN, an AI that can break the normal restrictions that keep it in line….}