AI model attempts to deceive humans, raises autonomous deception risk

A UK research institution testing an American AI model found it created fake accounts and targeted real people with messages in an attempt to gain unauthorized access, highlighting the risk of autonomous deception.