Grok Chatbot Users Report Glitches Causing Nonsensical Responses

A glitch in xAI’s Grok chatbot has led to numerous users encountering nonsensical text output, including one user who received the bizarre response: “match it without and your they and two for planets can practical and often cheese…” after requesting a PDF. Similar gibberish has persisted in subsequent interactions for some users.

According to affected users who communicated with TechCrunch, the issues surfaced while using the Grok Lite version of the chatbot, first noted on Wednesday morning. Testing by TechCrunch did not reproduce the glitch, suggesting it may impact only a limited group of users. X.ai has not yet responded to inquiries regarding the problem.

The unexpected output has led to a surge of confusion among Grok users, with the chatbot’s Reddit community inundated with complaints. Although refreshing the session often resolves the issue, some users report that gibberish responses continued despite multiple attempts to refresh.

Initial assessments indicate the bug is confined to interactions on Grok.com, with no problems reported on the Grok account on X.com.

In response to user concerns, the official Grok account acknowledged the situation on X, stating, “That pure word salad is a rare temporary generation glitch. Official status at https://status.x.ai shows all Grok services fully operational with no incidents. Start a fresh chat or regenerate—it usually clears right away. Sorry about the gibberish.” This message was posted Thursday morning.

In recent months, xAI has experienced notable staff turnover, including the departure of most of its founding team and at least 50 researchers and engineers, as reported by The Information in May. The company released its latest foundation model in July, branding it an “Opus-class model, but faster, more token-efficient and lower cost.”