Grok keeps sending gibberish responses to users

https://techcrunch.com/wp-content/uploads/2025/07/GettyImages-2218892225.jpg?resize=1200,804

10:32 AM PDT · August 20, 2026

An unusual glitch has resulted in the xAI’s Grok chatbot speaking gibberish to many users.

After asking the model to generate a PDF, one user received the response: “match it without and your they and two for planets can practical and often cheese…” with similar nonsense continuing for several paragraphs.

Checking source links, another user found a string of links to reinforcement learning research sites.

Affected users who spoke to TechCrunch said they were using Grok Lite, and noticed the issues as early as Wednesday morning. TechCrunch was unable to reproduce the issue in our own testing, and it is likely affecting only a small subset of users. X.ai did not respond to a request for comment.

Still, the result has been a lot of confused Grok users. The chatbot’s Reddit community has been overwhelmedwithcomplaints. Refreshing the session often restores...

Copyright of this story solely belongs to techcrunch.com. To see the full text click HERE

Read more

https://www.politico.com/dims4/default/resize/1200/quality/90/format/jpg?url=https%3A%2F%2Fstatic.politico.com%2Ff6%2F0a%2F4e472ed54f338f0590b872d39a4c%2Fgettyimages-2298703502.jpg

How a small, wealthy AI-safety network that organized in Signal chats grew into a coalition that beat the $100M super PAC Leading the Future in the midterms

Featured Podcasts Big Technology Podcast: AI Stockpocalypse, Anthropic Nears RSI?, Prediction Markets vs. Pollsters The Big Technology Podcast takes you behind the scenes in the tech world featuring interviews with plugged-in insiders and outside agitators. Subscribe to Big Technology Podcast. Machine Gods: The Purple Tsunami Against Data Centers The official