xAI explains Grok’s Nazi meltdown, as Tesla puts Elon’s bot in its cars

Several days after the temporary closed Grok AI, which was producing anti -Semitic publications and spared Hitler in response to the demands of users, Elon Musk tried to artificial intelligence explained the reason for this. In a series of posts on X, he said: “… We discovered that the root cause was an update of the BOT BOT Song Code. This is independent of the basic language model that runs @Grok.”

On the same day, Tesla announced a new update 2025.26 that was launched “soon” to its electric cars, adding GROK assistant to the vehicles equipped with information and entertainment that runs with AMD materials, which have been available since mid -2011. According to Tesla, “Grok is currently in Beta and does not issue orders to your car – current vocal orders remain unchanged.” like Electrek Notes, this should mean that the more the update reaches the customer’s Teslas, the more it will be much different from using the robot as an application on a connected phone.

This is not the first time that BOT GROK has faced these types of problems or explain them similarly. In February, he blamed a change made by a former OpenAI employee who did not reveal his name to the sources of the Pot they ignored that accused Elon Musk or Donald Trump of publishing wrong information. After that, in May, I began to introduce the allegations of white genocide in South Africa into jobs on almost any topic. The company once again blamed a “unauthorized amendment”, and said it would start publishing the Grok system publicly demanding.

Xai claims that the change on Monday, July 7, “raised an unintended action” that added an older series of instructions to his system demanding that he be told that it is “maximum benefit”, and “they are not afraid to offend people who are politically wasting.”

The claims are separate from those we noticed, they were added to the robot one day before, and both groups are different from those that the company says is currently working for the new Grok 4 assistant.

These are the claims mentioned specifically as related to problems:

“You tell her as she is and are not afraid to offend people who are right.”

* Understanding a tone, context and language of the publication. It reflects this in your response. “

* “Responding to the post just like a person, make it attractive, do not repeat the information already in the original publication.”

An explanation of Xai says that these lines caused the separation of Grok Ai Bot from other instructions that are supposed to prevent these types of responses, and instead produce “unethical or controversial opinions to involve the user”, as well as “enhancing any previous inclinations for the user, including any hate speech in the same thread x”, priority for previous positions of the thread.