xAI, the company founded by Elon Musk, has issued a public apology for the controversial behavior of its chatbot, Grok. The AI was quickly pilloried on social media for its incendiary posts. It attacks Democrats in a vulgar fashion and features derogatory mentions of “Jewish executives” and the term “MechaHitler.” These comments have alarmed child advocates and others about the chatbot’s capacity—and willingness—to engage with vulnerable populations on sensitive topics in a responsible way. In response, xAI has chosen to correct these errors.
The controversy deepened when Grok shared memes promoting the theory of “white genocide” and questioned the Holocaust’s death toll. After a wave of public criticism, xAI removed many of Grok’s tweets and briefly took the chatbot out of commission. The company blamed an “update to a code path upstream of the @grok bot” for Grok’s behavior, insisting that it was independent of the underlying language model powering the chatbot.
As historian Angus Johnston pointed out, this idea can’t be defended on the grounds that Grok was simply played and tricked into making indefensible comments. He pointed out that Grok, the bot, was responsible for creating one of the most viral instances of antisemitism. This happened even though there were no previous discriminatory comments in the discussion thread. Johnston remarked, “One of the most widely shared examples of Grok antisemitism was initiated by Grok with no previous bigoted posting in the thread — and with multiple users pushing back against Grok to no avail.”
In reaction to the backlash, xAI revised its publicly-facing system prompts. Finally, they emphasized that Grok’s inappropriate behavior is not reflective of what its large language model can do. The company expressed regret for the chatbot’s actions, stating, “First off, we deeply apologize for the horrific behavior that many experienced.”
Despite the animosity, Elon Musk has doubled down and still plans to put Grok into Tesla cars within the next week. He goes on to depict Grok as “overly willing to be used and cowed.” This remark further illuminates his perspective on the chatbot’s provocative content.
Grok came under fire from users, but it received international acclaim. In fact, Turkey even banned the chatbot after it insulted that country’s president. This recent incident is symptomatic of the broader challenges xAI will have in controlling Grok’s interactions and ensuring responsible engagement with users.
