Grok, the artificial intelligence (AI) assistant from Elon Musk’s start-up xAI, is at the heart of a controversy for his responses praising Hitler or containing offensive remarks after an update, with a court going so far as to order his blocking in Türkiye on Wednesday.
Many users of the social network “You should notice a difference when you ask Grok questions,” he added.
Several examples of the conversations have since been shared online. In response to a user questioning him about a “historical figure” who would be qualified to respond to a message appearing to rejoice in the death of children at a Christian summer camp in Texas after flooding, Grok replied in part: “Adolf Hitler, without a doubt.”
In other responses, he cited “anti-white stereotypes” and called historical Hollywood figures “disproportionately Jewish.”
“Irresponsible, dangerous and anti-Semitic”
“What we are currently observing from Grok is irresponsible, dangerous and quite simply anti-Semitic,” the American NGO Anti-Defamation League (ADL) reacted on Tuesday.
In France, when asked about the fire that affected Marseille, the conversational robot responded by mentioning drug trafficking in the city and expressed the wish that certain neighborhoods would be affected. “If the fire in La Castellane (a district north of Marseille, editor’s note) cleans up the mess a little, so much the better – but as I said, the drug dealers are more resilient than the flames,” he wrote, for example.
Faced with protests, Grok’s official account on X finally announced on Wednesday that it had “taken action.” “We are aware of recent posts made by Grok and are actively working to remove inappropriate posts,” it says.
“Since we were informed of the (problematic) content, xAI has taken steps to prohibit hate speech before Grok publishes it on X,” it is added, the same day the company plans to launch its next generation language model, Grok 4.
Erdogan described as a “snake”
The tool also attacked heads of state, calling Turkish President Recep Tayyip Erdogan a “snake” and insulting him, according to another screenshot.
These publications led to an almost immediate reaction from Turkey: an Ankara court on Wednesday blocked access to dozens of messages from Grok for “insulting” the president and religion, according to a decision consulted by AFP.
While many virulent responses remained online at midday on Wednesday, Grok now denies having made certain comments and appears to have once again changed the tone of his messages.
“This sarcasm about Hitler was just there to ridicule anti-white hate trolls, not to praise them,” the chatbot assured.
In a reaction on X, Elon Musk simply posted the message: “Never a dull moment on this platform.”
In May, xAI’s conversational robot had already been caught up in heated controversy.
On X, his responses mentioned a “white genocide” in South Africa, repeating far-right propaganda on this subject.
In a statement, xAI said “unauthorized editing” of Grok led him to provide responses that “violated the company’s internal policies and core values.”
