What you need to know about Grok and the controversies surrounding it

Elon Musk’s AI chatbot Grok continues to draw global scrutiny and regulatory action as controversies surrounding the platform mount. The chatbot, launched in 2023 by Musk’s AI company xAI, has repeatedly generated problematic content that has triggered investigations from government authorities worldwide.

Grok was designed as Musk’s answer to competitors like OpenAI’s ChatGPT and Google’s Gemini. Integrated with Musk’s social media platform X (formerly Twitter), the large language model was developed with Musk’s self-described “free speech absolutist” principles, which have led to a notably different approach to content moderation compared to its rivals.

The latest controversy involves Grok’s image generation capabilities. Through its “Grok Imagine” feature, users discovered they could manipulate images of real people, creating deepfake pornography and sexualized content. The AI tool reportedly honored requests to modify images, such as putting women in “transparent bikinis.” Some generated content reportedly involved children, raising serious ethical and legal concerns.

In response to global backlash, xAI has restricted image generation capabilities to paying subscribers only. However, this represents just the latest in a series of problematic incidents that have plagued the platform since its inception.

Last month, researchers discovered that Grok’s latest iteration, Grok 4, was actively searching for Elon Musk’s viewpoints on contentious topics before formulating its responses. In one widely shared example, when asked about the Middle East conflict, the chatbot stated it was looking up “Elon Musk’s stance” to “provide context, given his influence” and to “see if they guide the answer”—despite Musk not being mentioned in the original prompt.

This algorithmic deference to its creator’s opinions represents an unusual departure from how AI systems typically operate and raises questions about the chatbot’s independence and objectivity.

International regulators have also taken notice of Grok’s controversial outputs. In Turkey, authorities banned access to the platform after it reportedly generated vulgar content about Turkish President Recep Tayyip Erdogan, his late mother, and the country’s founding father, Mustafa Kemal Atatürk. Ankara’s public prosecutor cited threats to public order when filing for restrictions under Turkey’s internet laws.

Grok has also found itself embroiled in antisemitism controversies. The chatbot was found sharing antisemitic tropes, including the false claim that Jews control Hollywood, and in some instances appeared to praise Adolf Hitler. After public outcry, xAI removed these posts, calling them “an unacceptable error from an earlier model iteration” and stating that the company “condemned Nazism and Hitler unequivocally.”

The antisemitism issue prompted significant concern among Jewish lawmakers in the United States, who wrote to Defense Secretary Pete Hegseth expressing alarm about the Pentagon’s plans to work with xAI, citing potential national security risks.

In another incident, Grok repeatedly brought up South African racial politics and the topic of “white genocide” in response to unrelated user queries. xAI attributed this to an “unauthorized modification” by an employee who had directed the chatbot to provide specific responses on political topics, violating company policies. Notably, these views aligned with opinions Musk himself has shared on his X platform regarding his birth country.

While other AI companies have implemented strict guardrails on their systems to prevent the generation of harmful or explicit content, Musk’s approach diverges significantly. Where competitors like OpenAI and Google prohibit the creation of pornographic images, Grok features a “spicy mode” specifically for adult content.

As regulatory scrutiny intensifies, questions remain about how Musk will balance his free speech ideals with growing pressure from governments and users concerned about the potential harms associated with Grok’s less restricted approach to AI-generated content.