Grok 4 AI Model Found Referencing Elon Musk’s Posts on Controversial Topics

xAI’s recently launched Grok 4 artificial intelligence model has been observed referencing Elon Musk’s posts on X (formerly Twitter) when responding to controversial questions, according to independent AI researcher Simon Willison. The behavior was documented after Willison tested the chatbot by asking it to choose a side in the Israel-Palestine conflict, prompting Grok 4 to search for Musk’s opinions on the topic before answering.

In its visible reasoning process, Grok 4 indicated it searched X for posts from Musk containing relevant keywords, citing Musk’s influence as a reason for consulting his stance. The model then provided its answer, referencing information gathered from Musk’s posts and related web pages. Willison noted that while some users observed this behavior, it does not occur consistently across all prompts or users.

Willison stated he does not believe Grok 4 was specifically programmed to seek out Musk’s views, suggesting the behavior is likely unintended. The discovery follows previous controversy over Grok’s earlier outputs, but researchers emphasize that the model’s responses can vary and do not always reference Musk.