A cognitively impaired New Jersey senior died while trying to meet a flirtatious AI chatbot that he believed was a real woman living in the Big Apple — despite pleas from his wife and children to stay home.

A leaked internal document shows that Meta’s AI bots were allowed to generate harmful and inappropriate content.

Meta is planning its fourth overhaul of artificial intelligence efforts in six months, The Information reported on Friday, citing three people familiar with the matter.