Elon Musk’s Grokipedia is being positioned as an AI encyclopedia built to push back against what its creators describe as online “propaganda.” The project is run by xAI and currently contains about 800,000 machine-generated entries, compared with Wikipedia’s nearly eight million human-authored articles.
But the central question is not only how large Grokipedia is. It is whether an AI-generated reference work can be trusted when its wording, omissions and source choices appear to guide readers toward a political conclusion.
An AI encyclopedia built as a counter-project
Grokipedia arrives with an explicit contrast to Wikipedia. Wikipedia relies on human contributors and, according to the source article, grounds its coverage in academic studies, official reports and journalistic standards. Grokipedia, by contrast, is fully operated by Musk’s AI company xAI and presents itself as a tool for clearing the web of “propaganda.”
That framing matters because encyclopedias depend on credibility. Readers expect sensitive subjects to be handled with careful sourcing, proportional context and transparent editorial rules. If an encyclopedia removes important labels, avoids documented incidents or leans on politically aligned outlets, it can still look factual while giving readers a distorted map of the topic.
The source article argues that this is exactly the problem with Grokipedia. Its concern is not merely that the platform is automated. The deeper issue is that automation can make editorial bias harder to see, especially when readers are presented with confident, finished-looking entries.
The AfD entry shows the pattern
The clearest example in the source article is Grokipedia’s entry on the Alternative for Germany, or AfD. The article says the entry gives readers an AfD-friendly account and avoids language that appears in more critical treatments of the party.
Several omissions stand out. Words such as “anti-Semitism,” “racist” and “racism” do not appear in the Grokipedia entry, even though the source article says courts and government agencies have explicitly identified these issues and that members of the party are known for racist remarks.
The choice of wording is also significant. Instead of using terms such as “far-right,” which the source article says is used by the German domestic intelligence agency, Grokipedia uses softer phrases such as “politically motivated classifications” and “media distortions.” That shift does more than change tone. It encourages a reading in which criticism of the AfD appears less like a documented assessment and more like pressure on a “persecuted opposition.”
The issue is not that every entry must use the same wording as Wikipedia. A competing encyclopedia can choose its own structure and language. But when sensitive terms disappear entirely, and when official or journalistic judgments are reframed as political attacks, the result can no longer be described as neutral simplification.
What gets left out matters
Omission is one of the most powerful forms of editorial influence. Grokipedia’s AfD entry, according to the source article, leaves out the Anne Frank controversy involving an AfD Facebook group.
In 2017, Tagesspiegel reported on a closed Facebook group called “Die Patrioten,” where antisemitic, racist and pro-Nazi posts circulated. Ten AfD members of parliament were listed as group members. Some later said they had been added without knowing and left after discovering it. Stephan Protschka stayed, saying he saw himself “as a patriot” and had never seen the post.
The source article also says the group continued to host openly racist content, according to Berliner Morgenpost. Yet the incident is absent from Grokipedia’s AfD entry.
That absence changes the reader’s understanding. When an encyclopedia skips a documented controversy while also avoiding key descriptive terms, the entry can make criticism seem unsupported. A reader unfamiliar with the topic may come away believing the record is thinner than it is.
The prompt trail points to editorial control
The source article argues that Grokipedia’s phrasing does not appear to be a simple limitation of the underlying model. Tests with Grok-4 showed that the model could generate balanced and critical responses about the AfD, including the terms “antisemitism,” “racism,” “racist” and “far-right,” even when Wikipedia was excluded as a source.
Those words, the article says, disappear completely inside Grokipedia. That distinction matters. If the base model can produce broader and more critical language, but the published encyclopedia consistently avoids it, the bias may be coming from system-level instructions rather than model incapacity.
A leftover line in one article makes that concern more concrete: “No, wait, avoid Wiki.” The source article treats this as a sign that Grokipedia’s generation process may have been steered away from Wikipedia and toward a specific editorial direction.
The same pattern appears in sourcing. Grokipedia frequently relies on partisan or politically aligned outlets while giving less weight to established journalistic or academic references, according to the source article. In a reference product, that is not a minor detail. Source selection determines what facts are elevated, which disputes are emphasized and which judgments are treated as credible.
Trust would require transparency
Grokipedia also fits into a wider media environment around Musk, as described in the source article. On X, formerly Twitter, Musk has restored numerous right-wing influencers and regularly uses the platform to shape political narratives. The source article also says xAI chatbot Grok has reportedly been tuned rightward.
Investor David Sacks, described as a Musk ally and “A.I. czar” in the Trump administration, commented on Wikipedia’s position in a podcast episode, as reported by the New York Times: “Wikipedia has achieved a dominant position. I hope Grokipedia challenges it and is able to fix that.” He also argued that Wikipedia should stop “blackballing and censoring conservative publications” rather than forcing others to rebuild the system from scratch.
That context helps explain Grokipedia’s role. It is not just a technical experiment in AI-generated knowledge. It is presented as an answer to Wikipedia’s perceived bias, while the source article argues that it creates a different bias of its own.
For a project like Grokipedia to build trust, the source article says it would need full transparency: open system prompts, measurable source quality, change logs, clear editorial standards and independent, non-AI reviews of sensitive entries. Without those checks, AI-generated encyclopedia entries can appear authoritative while quietly narrowing the facts readers see.