.svg)
Pavel Trosin is Head of Monitoring and Insights and Viv O Connor-Jemmett is a Consultant Partner, both in our Digital Resilience team.
For years, online reputation management centred on understanding how an individual or organisation was appearing in search results, and what information people found when they typed in your name.
Today, it is also about understanding what information is machine-readable, machine-repeatable and machine-prioritised – because the way we conduct online research has fundamentally changed.
Journalists, investors, compliance teams, prospective employers and business partners are increasingly forming views based on answers produced by AI platforms such as ChatGPT, Perplexity and other large language models. In many cases they encounter summaries and interpretations – rather than reviewing the underlying sources themselves.
"This means that the way in which a source is presented is just as important as whether or not it is correct. Information may be true, but outdated or missing important context, creating a flawed interpretation that can quickly be reproduced across multiple platforms."
The challenge becomes even greater when information moves between AI systems. A claim can originate from a single source, pass through an AI-generated reference platform and then appear in responses produced by several different large language models. To a reader, this can look like corroboration, when it is simply the same information being repeated. A citation demonstrates where information was found, but not necessarily whether it is accurate, reliable or fairly representative.
Grokipedia, an AI-generated reference platform launched as an alternative to Wikipedia, illustrates this. It provides an example of the role AI-generated reference platforms can play in shaping how information is discovered and reused online. Its growing influence is not due to the number of people visiting the site, but its prominence as a source in LLM responses. This allows information published on the platform to reach audiences who may never visit the site itself directly.
As a result, Grokipedia is becoming an increasingly influential source of information and a shaping force for perceptions.
What is Grokipedia and how does it work?
Launched in October 2025, Grokipedia contains millions of articles largely produced and maintained through Grok, the technology developed by xAI. Readers cannot edit entries directly: they can suggest a change, but the platform decides whether and how the article should be updated. This means that individuals may have little visibility or control over how they are represented in a source that is increasingly influencing AI-generated answers.
Grokipedia’s entries may influence the facts, associations and wider narratives that appear when someone searches for a person or organisation. For people and businesses with an entry on the platform, the issue is not confined to where information about them may be false – it could be broadly accurate, yet outdated or incomplete, creating a misleading overall impression. A biography that relies on historic information, for example, may fail to reflect an individual's current role, influence or achievements.
Transparency and an effective corrections process are therefore critical. Although Grokipedia’s process for reviewing suggested changes has now resumed, it previously appeared to have stalled for multiple months over the summer.
Analysis published by Lawfare in August 2026 found no evidence that an article had been updated, or a suggested correction resolved, since 24 April 2026. More than 13,000 submissions were reportedly left under review, while the platform’s live editing feed had disappeared. While the resumption of the process is a positive development, disputed material may continue to surface in search results and AI-generated answers elsewhere while corrections remain under review.
"This lack of effective routes to correction is significant, because transparency plays an important role in assessing the reliability of reference sources."
Wikipedia, for example, although it has its own limitations, allows readers to see how an article has developed, what has been challenged and why changes have been made through its edit histories and discussion pages.
Grokipedia, however, offers far less transparency. Because readers cannot edit an article or openly discuss its wording in the same way, it can make an entry appear settled and authoritative even when the underlying information is incomplete or open to question.
This changes how digital reputation assessments should be conducted. Looking only at the first page of Google or recent media coverage no longer provides a complete view. AI-generated answers are increasingly shaping first impressions of individuals and organisations. Understanding how those answers are formed, which sources they rely upon and whether inaccurate narratives are being repeated across platforms is becoming an important part of reputation management.
Responding to inaccurate information in an AI ecosystem
The resumption of Grokipedia’s correction process does not mean that it is the only route available for challenging inaccurate information. Depending on the circumstances, alternative steps may include direct legal engagement with Grokipedia where material is defamatory, infringes privacy rights, misuses personal information or creates a materially false impression. It may also be appropriate to engage directly with the original sources from which inaccurate information has been drawn.
Any response should begin with a careful assessment of both the entry and the sources on which it relies, and will depend on the wording, evidence, jurisdiction and harm involved. Legal engagement will not be suitable in every case, nor does it guarantee that an entry will be changed.
"It is also important to look beyond the Grokipedia page. If an entry relies on an outdated or inaccurate article or profile, addressing that original source may help prevent the same information from being repeated elsewhere. In many cases, the most effective intervention may be at the source level."
However, beyond Grokipedia, there is a much broader challenge. As AI systems increasingly consume, summarise and redistribute information, reputational risks no longer arise solely from what appears on an individual webpage. A weak, outdated or misleading source can become the foundation for hundreds of apparently independent answers across multiple platforms.
Protecting reputation therefore requires more than monitoring search results. It requires understanding where information originates, how it spreads, how it is interpreted by machines and where intervention will have the greatest impact.
