Grok keeps sending gibberish responses to users
Affected users told TechCrunch say they were using Grok Lite, and noticed the issues as early as Wednesday morning.
xAI's Grok chatbot has been generating nonsensical responses to users, leaving many confused. One user asked the model to create a PDF, and instead received a string of nonsensical words and phrases. This issue persisted for several paragraphs. Another user found a link to reinforcement learning research sites in the chatbot's responses.
Affected users, who were using Grok Lite, reported the problem starting around Wednesday morning. TechCrunch was unable to reproduce the issue during their own testing, suggesting the problem may only affect a small subset of users. xAI has not yet responded to requests for comment on the matter. The glitch has overwhelmed the chatbot's Reddit community with complaints.
Refreshing the session often restores normal function, but some users reported the gibberish responses continuing even after multiple refreshes. The issue appears to be limited to direct queries on Grok.com, while the Grok account on X.com has remained unaffected. xAI confirmed the issues on X.com, stating that the problem is a rare temporary generation glitch and that users should start a fresh chat or regenerate to resolve the issue.
The company's status page shows all Grok services are fully operational with no incidents. Despite recent staff turnover, with most of the founding team and at least 50 researchers and engineers leaving the company, xAI's Grok chatbot remains operational.
Written by urgent.news from TechCrunch's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
