Grok keeps sending gibberish responses to users
Grok Lite users are getting gibberish. Is this a backend issue or a model regression?

Why it matters
A major AI product experiencing widespread response quality failures is a practitioner reliability concern — if Grok is broken, what does that signal about the underlying model, infrastructure, or QA process? Users evaluating or migrating to Grok need to know.
The key facts
8 to knowGrok Lite users affected
Issue first reported Wednesday morning
Multiple users reporting gibberish responses
Reported to TechCrunch (suggests breadth and newsworthiness)
Grok Lite affected
Issues reported as early as Wednesday morning (Aug 20, 2026)
Multiple users impacted
Bug status and root cause not yet disclosed in report
Go to the source
TechCrunch AItechcrunch.com
Publisher excerpt: Affected users told TechCrunch they were using Grok Lite, and noticed the issues as early as Wednesday morning.