Grok censors Musk, Trump search results; company blames ex-OpenAI employee
What's the story
Grok, Elon Musk's ChatGPT competitor, has recently been embroiled in controversy for temporarily blocking search results Elon Musk and Donald Trump. The problem was first observed by users who discovered that the AI chatbot was programmed not to pull up sources accusing either of them of spreading misinformation. The bizarre behavior was later traced to an unauthorized update by a former OpenAI employee now at xAI, Igor Babuschkin, xAI's head of engineering, revealed.
System prompt
Unauthorized update to Grok's system prompt caused the issue
Babuschkin clarified that the problem stemmed from an unauthorized change in Grok's system prompt, the internal rules that determine how an AI responds to queries. He said, "An employee pushed the change because they thought it would help, but this is obviously not in line with our values." "I haven't thrown anyone under the bus. You were making a point about OpenAI's culture vs xAI's culture. I've given you more details on what happened...," replied Babuschkin to a user's comment.
Twitter Post
The employee's update was not authorized by xAI boss
Wish they would have talked to me or asked for confirmation before pushing the change.
— Igor Babuschkin (@ibab) February 23, 2025
Mission statement
Grok's mission and recent controversial statements
Musk describes Grok as a "maximally truth-seeking" AI, whose mission is to "understand the universe." However, since the release of its latest model, Grok-3, it has made some controversial statements. The chatbot claimed President Trump, Musk himself, and Vice President JD Vance are "doing the most harm to America." Additionally, it suggested both Musk and Trump deserve capital punishment.