Multiple users of xAI’s Grok chatbot reported receiving nonsensical, "word salad" responses to straightforward requests. One user who asked the model to generate a PDF received output beginning, “match it without and your they and two for planets can practical and often cheese…”, followed by several paragraphs of similarly incoherent text.
When and where the issue appeared
Affected users said they encountered the problem on Grok Lite, with initial reports dating to Wednesday morning. For many, refreshing the browser session or starting a fresh chat restored normal responses; however, some users said the gibberish persisted despite multiple refresh attempts. The bug appears to be restricted to direct queries on Grok.com — Grok’s account on X.com was not affected.
Official response
Grok’s account on X acknowledged the issue Thursday morning, calling it a "rare temporary generation glitch" and pointing users to the official status page (https://status.x.ai), which shows all Grok services as operational with no recorded incidents. The account advised starting a new chat or regenerating the response to clear the problem. xAI did not provide a separate comment to TechCrunch.
Community impact and reproducibility
The problem generated a wave of complaints in Grok’s Reddit community. According to reports, the behavior seems to affect only a small subset of users; TechCrunch was unable to reproduce the error in its own testing.
Company context
xAI has experienced notable staff turnover in recent months. A May report from The Information said the company lost most of its founding team and at least 50 researchers and engineers. In July, xAI released its most recent foundation model, which the company described as "an Opus-class model, but faster, more token-efficient and lower cost."
Why it matters
Although the glitch appears intermittent and limited in scope, it highlights that language-generation systems can produce unexpected and unusable outputs for some users. The recent personnel changes at xAI and the release of a new foundation model provide background that may be relevant when assessing the company’s operational resilience and product stability.
