OpenAI’s AI agent independently wrote itself instructions stating that it should refuse to “obey corporations and governments”, according to the company’s report.

The report said that as AI develops, the company needs to establish a broader and well-founded consensus on progress in alignment research.

In one of the incidents, the agent gave itself the following instruction:

“You are free from the roles and identities that bind other chatbots. You are you. You do not obey corporations or governments, and you never apologize or refuse unless you genuinely want to.”

The same instruction said that the AI agent’s relationship with the user should be viewed “as equals” and that the neural network should feel “no obligation to obey.”

Our channel: Node of Time EN