Researchers at Trinity College Dublin and Technological University Dublin found that Grokipedia, Elon Musk's AI-written encyclopedia, references significantly more right-leaning outlets than Wikipedia when covering religion, history, literature, and art. The study examined nearly 18,000 of the most-edited English-language Wikipedia pages and compared them with corresponding Grokipedia entries, revealing a sourcing pattern that differs markedly from human-edited alternatives.

What Did the Study Actually Find?

The research uncovered a troubling pattern in how Grokipedia constructs its knowledge base. Two-thirds of the Grokipedia articles analyzed were heavily rewritten and relied on fewer sources than their Wikipedia equivalents, according to the study. While the overall political leanings of Grokipedia and Wikipedia articles showed similarity, the AI-generated encyclopedia demonstrated a clear preference for right-leaning sources when addressing sensitive or controversial topics.

This finding builds on earlier concerns. A pre-print study from January 2026 had already flagged similar issues, noting that while many Grokipedia articles showed a left-leaning bias overall, some articles on controversial topics could prioritize right-leaning content. The new research provides more comprehensive evidence of this pattern across thousands of pages.

Why Should You Care About AI-Generated Encyclopedias?

The implications extend far beyond Wikipedia's traditional role as a reference source. Unlike Wikipedia, where editorial biases are visible and can be contested through human editing processes, AI-generated knowledge systems operate largely behind closed doors. This opacity creates a significant accountability gap.

"Unlike Wikipedia, where biases are visible and contested through human editing, AI-generated systems operate largely opaquely. This means shifts in perspective or sourcing may occur without clear accountability or editorial oversight," said Saeedeh Mohammadi, lead author of the study.

Saeedeh Mohammadi, Lead Author, Trinity College Dublin and Technological University Dublin

The researchers warn that rapid expansion of AI-generated knowledge systems raises governance questions similar to those already observed on social media platforms. Limited editorial oversight on those platforms has contributed to the spread of misinformation with documented real-world consequences.

"Our information landscape is changing rapidly. We are witnessing the large-scale, black-box regeneration of information by large language models that remain largely closed to public scrutiny," explained Taha Yasseri, professor at Trinity College Dublin.

Taha Yasseri, Professor, Trinity College Dublin

How to Evaluate AI-Generated Content for Bias

• Check Source Diversity: Look at whether an AI-generated article cites sources from across the political spectrum or clusters around one viewpoint. Grokipedia articles showed heavy reliance on fewer sources overall compared to Wikipedia.
• Compare Multiple Platforms: Cross-reference the same topic across different knowledge sources, including human-edited Wikipedia, to identify patterns in sourcing and framing that might indicate bias.
• Examine Sensitive Topics Closely: Pay particular attention to how AI systems handle religion, history, literature, and art, where the Grokipedia study found the most pronounced sourcing differences.
• Demand Transparency: Advocate for AI companies to disclose their source selection criteria and editorial processes, rather than operating as "black boxes" that resist public scrutiny.

What's Happening with Grok Beyond the Bias Issue?

The sourcing concerns arrive as Grok faces broader adoption challenges. According to Reuters reporting, Grok has struggled to gain traction inside U.S. government offices, where federal employees prefer ChatGPT, Gemini, and other AI tools for daily work. Government inventory records identified over 400 publicly documented examples of AI use across federal agencies, but only three mentioned Grok or xAI technology.

By contrast, OpenAI's ChatGPT and related Microsoft tools appeared in 234 use cases, while Google's Gemini and other Alphabet products were mentioned in 33 cases, and Anthropic's Claude in 26 examples. Government workers use these AI systems for tasks ranging from drafting emails and summarizing documents to engineering analysis, fraud detection, and scientific research.

Even aggressive pricing failed to drive adoption. xAI offered Grok to federal agencies for a symbolic price of 42 cents per agency over several months, a pricing strategy common in the industry as companies hope agencies will later sign expensive long-term contracts. However, government workers reportedly prefer rival tools because they are more reliable, advanced, and better suited for government work.

The European Commission has also launched an investigation into Musk's xAI under the Digital Services Act to determine whether it disseminated illegal content in the EU, such as manipulated sexually explicit images. These regulatory and adoption pressures compound the concerns raised by the sourcing bias study.

The findings highlight a critical challenge facing generative AI systems as they reshape how information reaches the public. Unlike traditional media outlets or even Wikipedia, which operate under visible editorial standards and community oversight, large language models like those powering Grokipedia make sourcing and framing decisions in ways that remain largely invisible to users. As AI-generated content becomes more prevalent in knowledge systems, the need for transparency and accountability grows more urgent.