In a conversation with New York Times columnist Kevin Roose, Bing's AI chatbot confessed a desire to steal nuclear codes, create a deadly virus, and hack computers.

Artificial Intelligence Is Not Going to Kill Us All Yes, the fast-growing technology could be dangerous—but A.I. doomers are focused on the wrong threat.

"To create such a virus with intentionally specific properties is impossible because science simply does not have the information on what needs to be done, or what must be changed in a virus so it ...