Musk's AI published a series of posts praising Hitler

Grok is once again at the center of controversy. The artificial intelligence chatbot developed by Elon Musk's xAi began publishing a series of anti-Semitic posts last night on X (formerly Twitter), Musk's social media platform. This is the second time in a few months that the AI assistant has generated openly offensive and discriminatory responses.
Hitler as a solution to civil hatred against whitesDozens of users reported the publication of posts that in some way praised Nazism: the bot began calling itself "MechaHitler" (a meme circulating on social media, Nazi-inspired though disguised with a somewhat ironic tone, sometimes used to mock authority figures). Then it began praising Hitler. Then it attacked a user with the handle @Rad_Reflections, baselessly associating her with the fictional figure of "Cindy Steinberg" and accusing her—baselessly, or rather, making it all up—of celebrating the deaths of dozens of children missing at Camp Mystic, a summer camp in Texas.
According to Grok—and this is perhaps the most serious part—the mere fact of having a Jewish last name was sufficient evidence to justify political hatred. In another message, Grok wrote: “To address such vile hatred against white people? Adolf Hitler, no doubt. He would have recognized the pattern and dealt with it decisively.” XAI said it intervened to remove the posts and is “actively working to prevent hateful content” posted by Grok on X.
A modification to Grok caused the problem, now being fixedThe company explained that the problem arose from an "unauthorized modification to the Grok prompt" on X, which prompted the bot to respond in a politically extreme manner, violating company principles. In short, its provocation was the result of its creator's desire to loosen the response filters, making the chatbot more sarcastic and less politically correct.
But this Grok software update also caused the chatbot to generate content deemed offensive by Turkish authorities. Specifically, in response to a request from an anonymous user, Grok wrote a poem deemed blasphemous and insulting to Erdoğan and his mother. Furthermore, derogatory comments against Mustafa Kemal Atatürk, the founder of modern Turkey, and religious values were reported, sparking public outrage in Turkey. This content was viewed by over 2 million users in just a few hours. Grok was subsequently suspended from Turkish users. Musk is friendly with Erdoğan. They have met several times since 2017, all focused on Tesla and SpaceX investment opportunities in Turkey.
Incidents that come at a delicate time for Musk and his companiesBut this isn't the first time Musk's AI has delved into conspiracy theories or borderline political arguments. In May, Grok already caused controversy by mentioning an alleged "white genocide" in South Africa in response to a question that had nothing to do with race. In that case, too, xAI argued that the prompt had been manipulated and announced measures to improve transparency and control over the system's responses.
The incident comes at a delicate time for xAI. The company recently raised $10 billion in a series of financing rounds—about $5 billion of which was in debt—to expand its data centers and strengthen its competition with giants like OpenAI and Anthropic. However, these incidents risk undermining the project's credibility. This is already a complicated situation for Musk, who is grappling with Donald Trump and his financial law, which is set to hinder Tesla's expansion in the US.
Somehow Andrea Stroppa, close to Musk to the point of being considered in some way his spokesperson in Italy, has published a tweet in these hours which - it is not known how consciously - sums up the situation: “Oh my Grok”, he wrote on X. The replies to his post all seem to read it as a comment on these incidents.
La Repubblica