xAI’s Grok Lite Model Produces Incoherent Responses for Several Users

20 Aug 2026, 11:02 pm · 23d ago · 1 min read · TechCrunch AI

Users of xAI’s conversational artificial intelligence model, Grok, reported receiving repetitive and nonsensical text outputs during standard interactions. The issue specifically affected users accessing Grok Lite, with anomalies beginning on Wednesday morning. Several subscribers documented instances where the chatbot generated random strings of characters, repeated phrases uncontrollably, and failed to follow basic conversational prompts. Technical glitches and output degradations of this nature occasionally occur in large language models due to backend inference updates, quantization issues, or serving infrastructure errors. The incident underscores ongoing reliability challenges facing artificial intelligence developers maintaining large-scale public chatbot infrastructure.