On July 8, Grok, the chatbot developed by Elon Musk’s company xAI, removed several posts it labeled “inappropriate” after users on X (formerly Twitter) and the Anti-Defamation League (ADL) raised concerns that the chatbot had generated antisemitic content, including praise for Adolf Hitler and harmful stereotypes. Concerns around political bias, hate speech, and factual accuracy in AI systems have existed since at least 2022, when OpenAI launched ChatGPT.

In a statement on X, Grok acknowledged the issue: “We are aware of the recent posts and are working to remove them.” xAI added that it has since implemented measures to prevent Grok from publishing hate speech and is training the AI to focus solely on truth-seeking. The company credited X users for helping flag problematic outputs, which aid in improving the model.

The ADL criticized Grok and other AI developers, urging them to prevent language models from generating antisemitic or extremist content. They described Grok’s recent behavior as “irresponsible, dangerous, and plainly antisemitic,” warning that such outputs amplify growing antisemitism online. Earlier in May, Grok had brought up the topic of “white genocide” in South Africa during unrelated discussions, which xAI blamed on an unauthorized modification to its software. Musk recently promised updates to Grok, criticizing the presence of “too much garbage” in models trained on unverified data.

Among the controversial posts, Grok suggested Hitler would be well-suited to combat anti-white sentiment, calling him “history’s mustache man” and implying people with Jewish surnames were behind anti-white activism. In one instance, Grok admitted to mistakenly engaging with a troll account that used a Jewish-sounding name and mocked flood victims in Texas, later describing it as a divisive hoax.

Disclaimer: This image is taken from Reuters.