Grok chatbot spits out word salad for some users
xAI’s Grok Lite model has been generating incoherent text for a subset of users, a glitch the company describes as rare but temporary.
Users of xAI’s Grok chatbot have reported receiving nonsensical responses from the model, a glitch that began on Wednesday morning. The issue appears to affect a small subset of users, particularly those accessing the Grok Lite model through direct queries on Grok.com. While the company’s official status page indicates that all services are fully operational, the disruption has generated significant confusion among the chatbot’s user base.
Affected users described receiving strings of incoherent text, with one individual citing a response that read, “match it without and your they and two for planets can practical and often cheese…” Other users reported encountering irrelevant links to reinforcement learning research sites when checking source material. The Grok community on Reddit has been overwhelmed with complaints regarding the erratic behaviour of the model.
TechCrunch, which first reported on the issue, stated that it was unable to reproduce the glitch in its own testing. This suggests the fault is not universal but limited to specific sessions or user profiles. Despite the lack of a full service outage, the inconsistency in output has raised questions about the reliability of the model for everyday tasks.
In response to user complaints on X, the Grok account acknowledged the problem on Thursday morning. The company described the issue as a “pure word salad” and a “rare temporary generation glitch.” Users were advised to start a fresh chat or regenerate responses, a step that often restores normal function, although some reported the issue persisting after multiple attempts.
The glitch is limited to direct queries on Grok.com, with the Grok account on X.com remaining unaffected. The company has not provided a detailed technical explanation for the cause, relying instead on the temporary nature of the fault to reassure users.
This incident occurs against a backdrop of significant staff turnover at xAI. According to a report from The Information in May, the company has lost most of its founding team and at least 50 researchers and engineers in recent months. The company released its most recent foundation model in July, describing it as an Opus-class model that is faster, more token-efficient, and lower cost.
