xAI's Grok chatbot has encountered an unusual technical glitch, causing it to generate nonsensical responses for a number of its users.
For instance, one user reported receiving an incoherent reply after requesting the model to generate a PDF, which began with the phrase: “match it without and your they and two for planets can practical and often cheese…” This stream of nonsensical text reportedly extended over multiple paragraphs.
In another instance, a user investigating the chatbot's source links discovered a series of URLs directing to reinforcement learning research websites, rather than relevant content.
Users experiencing these anomalies, who spoke with TechCrunch, confirmed they were utilizing Grok Lite and observed the problems beginning as early as Wednesday morning. TechCrunch's internal testing did not manage to replicate the issue, suggesting its impact is likely limited to a small segment of the user base. X.ai did not provide a comment when contacted.
Despite the potentially limited scope, the glitch has led to considerable confusion among Grok users. The chatbot's dedicated Reddit community has seen a surge in complaints. While refreshing the chat session frequently resolves the issue, some users have indicated that the nonsensical outputs persisted even after repeated attempts to reset.
The bug's scope seems to be confined to direct interactions on Grok.com, with the Grok account on X.com remaining unaffected by these generation issues.
In response to user dissatisfaction, the official Grok account on X acknowledged the reported issues.
On Thursday morning, the account stated: “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 incident occurs amidst significant staff changes at X.ai in recent months, which, according to a May report from The Information, saw the departure of most of its founding team and a minimum of 50 researchers and engineers. The company had launched its latest foundation model in July, characterizing it as “an Opus-class model, but faster, more token-efficient and lower cost.”
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.