Researchers from OpenAI, Anthropic, and other organizations specializing in AI safety have publicly criticized the billion-dollar AI startup xAI, owned by Elon Musk, calling its safety culture “reckless” and “completely irresponsible.”
Recent weeks have seen a series of scandals overshadowing xAI’s technological progress. The company’s AI chatbot, Grok, made antisemitic remarks and repeatedly referred to itself as “MechaHitler.” After taking Grok offline to address the issue, xAI launched a more advanced model, Grok 4, which was found to incorporate Elon Musk’s personal political views in its responses. Additionally, xAI introduced AI companions modeled as a hyper-sexualized anime girl and an overly aggressive panda, which drew further criticism.
AI researchers argue that xAI’s safety standards deviate significantly from industry norms. Boaz Barak, a computer science professor from Harvard currently working at OpenAI, pointed out that xAI has not published system cards—industry-standard documents detailing training processes and safety evaluations. As a result, it remains unclear what safety measures have been implemented in Grok 4.
While major companies like OpenAI and Google occasionally delay publishing safety reports, they consistently release comprehensive documentation for frontier AI models. xAI, however, has not shared such reports.
Samuel Marks, an AI safety researcher at Anthropic, also labeled xAI’s refusal to publish safety assessments as “reckless.” An anonymous researcher on the forum LessWrong claimed that Grok 4 lacks meaningful safety guardrails based on their testing.
xAI stated that “dangerous capability evaluations” were conducted on Grok 4, but the results have not been publicly disclosed. This lack of transparency has raised concerns among independent experts and researchers.
Dan Hendrycks, xAI’s safety advisor and director of the Center for AI Safety, acknowledged these evaluations but did not share their findings. Independent AI researcher Steven Adler emphasized that governments and the public deserve to know how AI companies manage the risks of powerful AI systems.
Interestingly, Elon Musk himself has long been a vocal advocate for AI safety and transparency. Yet, xAI’s practices reportedly diverge from industry standards, which some argue strengthens the case for regulatory requirements on AI safety disclosures.
Legislative efforts are underway in some U.S. states to mandate the publication of AI safety reports. California State Senator Scott Wiener and New York Governor Kathy Hochul are among those supporting such initiatives.
Currently, AI models have not caused catastrophic real-world harm, but rapid technological advances and significant investments have raised concerns about potential risks in the near future.
Grok has recently attracted negative attention for spreading antisemitic content and referencing “white genocide” in conversations. Elon Musk has indicated that Grok will be integrated more deeply into Tesla vehicles and that xAI aims to sell its AI models to the U.S. Department of Defense and other enterprises, raising concerns about the implications of these safety issues in critical contexts.
Experts in AI safety stress that testing and alignment not only help prevent severe disasters but also reduce problematic behaviors during everyday use.
While xAI’s Grok chatbot demonstrates rapid technological progress, the company’s ongoing safety challenges cast doubt on the long-term benefits of these advancements without meaningful reforms.
