Grok Spits Out Nonsense Text, xAI Blames Rare Generation Error
A chatbot that suddenly outputs gibberish, with xAI insisting services are normal, is the kind of trust break that lands hardest right after the company touted a faster, lower-cost foundation model.
Reporting from 1 source: GameBusiness.jp.
xAI's Grok chatbot began producing nonsensical responses this week, confusing users across X and Reddit. One user asked for a PDF and got endless meaningless text. Others saw lists of unrelated links. The issue appears on Grok.com, mainly for Grok Lite users, while the Grok account on X.com works. xAI's official account called it a rare, temporary generation error and said the status page shows all services normal.
The trouble started Wednesday and spread across X and the Grok Reddit community, with users trading workarounds. Refreshing the session often clears the bad output, but some report the gibberish persists through multiple refreshes.
The failure is isolated to direct access on Grok.com, mainly affecting Grok Lite users. No problems were confirmed on the Grok account inside X.com.
xAI's official account posted Thursday morning that the incoherent text is a rare, temporary generation error, and pointed users to the status page, which shows all Grok services operating normally. The company advised starting a new chat or regenerating.
The incident follows a rough stretch for the company. The Information reported in May that at least 50 researchers and engineers, including most of the founding team, had left in recent months. In July, xAI released its latest foundation model, calling it an Opus-class model that is faster, more token-efficient, and lower cost.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.
Sources
- GameBusiness.jp xAIのGrokが突如意味不明な回答を並べ、困惑するユーザー相次ぐ