xAI’s Grok chatbot began producing nonsensical, incoherent responses to some users in August 2026, with complaints surfacing as early as Wednesday morning. The issue appears limited to Grok Lite users accessing the chatbot directly on Grok.com.
One affected user asked Grok to generate a PDF and received the response: “match it without and your they and two for planets can practical and often cheese…” followed by several additional paragraphs of similar nonsense. Another user reported receiving a string of links to reinforcement learning research sites when checking source links.
The Grok account on X confirmed the problem Thursday morning, describing it as “a rare temporary generation glitch.” The account directed users to start a fresh chat or regenerate their query, noting that doing so “usually clears right away.” The official xAI status page at status.x.ai listed all Grok services as fully operational with no incidents reported.
Refreshing the session restored normal function for some users, but others reported that gibberish responses persisted through multiple refreshes. The chatbot’s Reddit community was flooded with complaints. TechCrunch was unable to reproduce the issue in its own testing, suggesting the bug is affecting only a subset of users. The Grok account on X.com was not affected. xAI did not respond to a request for comment.
The glitch arrives during a turbulent period for xAI. According to a May report from The Information, the company has experienced significant staff turnover in recent months, losing most of its founding team along with at least 50 researchers and engineers. xAI released its most recent foundation model in July, describing it as faster, more token-efficient, and lower cost than previous versions.
Source: TechCrunch