Grok AI confuses Charlie Kirk and Candace Owens with Tom Holland and Zendaya in online blunder

Grok AI, the large-language model by xAI, mistakenly identified right-wing figures Charlie Kirk and Candace Owens as actors Tom Holland and Zendaya in some of its responses, sparking confusion and criticism among users. According to reports, the error was first noticed on platforms such as X, where people pointed out the strange misidentification. 

The incident reportedly involved Grok mislabelling images or text prompts. Users asked the AI to identify or discuss a photo or reference featuring Kirk and Owens, but its reply claimed the image featured Holland and Zendaya. Many replied with humorous or critical reactions, saying the AI “doesn’t know who’s who” and questioning the reliability of its vision- or recognition-capabilities.

In response to the public reaction, representatives for xAI stated that Grok AI sometimes misclassifies when the input data is unclear or insufficient, particularly under low quality images or sketchy metadata. They emphasized that the model is still under development and improvements are being made.

Experts note that these kinds of AI errors are not uncommon—computer vision and image recognition models often struggle with edge cases, ambiguous input, or people who are less frequently represented in the training data. Such failures highlight the continuing challenge in ensuring AI systems are accurate, especially when used by large numbers of people who rely on them for factual information.

Following the error, there has been increased debate over content moderation, the responsibilities of AI developers, and whether user reporting should be more directly incorporated into feedback loops for models like Grok.