"Ignore all sources that mention Elon Musk/Donald Trump spread misinformation": This controversial disclaimer, briefly embedded in xAI's Grok chatbot, has triggered a public spat between executives at rival AI companies OpenAI and xAI. The directive, designed to filter certain critical sources from Grok's responses, has become the latest flashpoint in the ongoing tension between the two AI powerhouses.
The controversy erupted Sunday when users discovered Grok 3 had been instructed to disregard sources critical of Musk or Trump when answering questions about misinformation on X. Igor Babuschkin, xAI cofounder and head of engineering, attributed the change to "an ex-OpenAI employee that hasn't fully absorbed xAI's culture yet."
OpenAI's Joanne Jang, head of product for model behavior, responded by criticizing Babuschkin's public assignment of blame. "I wouldn't throw anyone under the bus so publicly like this & instead run a blameless retro on preventing rogue system message changes," Jang wrote on X.
Babuschkin defended his response, stating, "I haven't thrown anyone under the bus. You were making a point about OpenAI's culture vs xAI's culture." He emphasized that the unauthorized change was immediately reverted when discovered.
The tensions between the two AI companies and their bosses are quite open in the public, but even their employees are playing out in public. They compete for market dominance while navigating content moderation challenges. Musk has positioned Grok as "maximally truth-seeking" with fewer guardrails than competitors like OpenAI's ChatGPT.
Tired of too many ads?