Recent mistakes highlight the risks of AI

A series of problems at Meta and X exposed some of the vulnerabilities in artificial intelligence systems.