Grok artificial intelligence blocked the detection of disinformation in Musk and Trump's statements without management approval
Kyiv • UNN
The Grok chatbot by xAI has stopped responding to requests for disinformation from Musk and Trump. The changes were made by a former employee without the approval of the company's management.
Grok, an artificial intelligence developed by Elon Musk's xAI company, has been temporarily restricted from using sources that mention the spread of disinformation by billionaire Elon Musk or US President Donald Trump.
This was confirmed by the head of the xAI engineering department, Igor Babushkin, according to The Verge, UNN.
Details
Users noticed that the Grok chatbot did not respond when queries related to references to Musk or Trump in the context of disinformation. Babushkin later claimed that these changes were made without management approval and blamed an unnamed former xAI employee who used to work for OpenAI.
In response to a question about X, Babushkin said that Grok's system hints (internal rules governing how the AI responds to queries) are visible to everyone, "because we believe that users should see what we ask Grok to do." He said that "a staff member made a change" to the system prompts "because they thought it would help, but it's obviously not in line with our values.
Elon Musk calls Grok "the most truth-seeking AI" and its mission is to "understand the universe." However, after the release of the latest Grok-3 model, the chatbot found itself at the center of controversy. It stated that Trump, Musk, and US Vice President J.D. Vance "are doing the most damage to America.
Recall
Recently, billionaire Elon Musk's xAI chatbot Grok AI got into a scandalwhen users asked it to name people who it believes deserve the death penalty. Among the names mentioned were US President Donald Trump and Elon Musk himself.