xAI’s Grok chatbot began spitting out gibberish this week in response to direct queries, TechCrunch reported. According to TechCrunch, one user reported that they asked Grok to generate a PDF, receiving a reply beginning “match it without and your they and two for planets can practical and often cheese…” (It just sort of went on from there, apparently.) Another user found that source links provided by Grok led to reinforcement learning research sites instead. Users told TechCrunch they first noticed the issues on Grok Lite, beginning on Wednesday morning. Complaints on the Grok subreddit included a reply to the word...
Monitoring Antisemitism Intel
