The AI giant’s safety chief said the model ‘didn’t quite meet the bar’ of the firm’s security standards. Read More
