Musk’s xai: unauthorized modification made Grok “ white genocide ”

Elon Musk‘S XAI said Thursday evening that his boot of response Grok kept talking about the “white genocide” in South Africa under unrelated positions due to an “unauthorized modification” on the backend.

“On May 14, around 3:15 pm PST, an unauthorized modification was made at the prompt of the Grok response bot on X,” XAI wrote on XA on Thursday.

“This change, which ordered Grok to provide a specific response on a political subject, violated the internal policies and the fundamental values ​​of XAI. We have conducted an in -depth investigation and implemented measures to improve the transparency and reliability of Grok,” added XAI.

XAI said that he would make several modifications to his code review policies. This includes the publication of the Grok system publicly invites GitHub so that users can provide their comments and introduce additional checks to their code examination process.

“We are setting up a 24/7 surveillance team to respond to incidents with Grok’s responses that are not captured by automated systems, so that we can respond more quickly if all other measures fail,” he added.

XAI did not immediately respond to a request for comments from Business Insider.

XAI’s Friday’s response comes after X people noticed that Grok did not stop giving unsolicited answers to the “white genocide” in South Africa under unrelated messages on Wednesday.

It is not known when the first articles of Grok on the “white genocide” were published. In his conversations with Business Insider, the response bot has given variable answers. He first said that he had been informed to do so by his “creators”, before saying in a subsequent question that his answers were guided by a “temporary bug”.