The Grok artificial-intelligence chatbot resumed responding to user posts on 15 July after being offline for about three days, according to messages on the social-media platform where the service operates. Developers said they had identified "a couple of problems" with the Grok 4 model, investigated them and implemented fixes before restoring the account’s ability to reply publicly. No further details about the nature of the glitches or their impact on user data were disclosed.