The Grok team chalked up the slew of inflammatory statements to a malfunctioning code update, not the tool's underlying AI model, and said the issue has now been resolved.