Listen to this article · 18 minutes
Read by an AI voice.
AI reputation management is the ongoing work of examining how AI systems describe an organisation, its brands and its leaders, then addressing the gaps between those descriptions and the public evidence. It means managing your reputation as represented in AI answers, rather than using AI software to manage reviews or sentiment, and it does not mean controlling what AI says.
The thing being managed is AI reputation (AIR): the picture someone receives when they ask an AI tool about a company, brand or individual. It might be a short introduction, an assessment of credibility, a comparison with competitors or an account of past controversies.
That picture can influence how a customer, investor, prospective employee or journalist understands the subject. Managing it requires looking beyond the organisation's own communications to the wider public record from which AI can draw information.
The three layers behind AI reputation
Adelbury's framework for AI reputation (AIR) groups the public evidence behind it into three layers: what you say, what you do and what others say about you.
The framework was developed in The AI Reputation Audit, published in September 2026. It provides a way to examine the information that can influence AI-generated descriptions and identify where an organisation's public record is consistent, incomplete or contradictory.
These are categories for analysing evidence, not a claim that every AI system sorts or weighs its sources in the same way.
What you say: declared narrative
The first layer is how an organisation describes itself.
Its website, company profiles, leadership biographies and published statements explain its identity, expertise, strategy and values. These are generally the sources over which the organisation has the greatest control.
They establish what the organisation wants people to understand about it. But a company's own description is a claim, not independent evidence that the claim is accurate.
Managing this layer means keeping information current, removing contradictions and ensuring important claims have credible supporting evidence. An outdated leadership biography or a website describing a business model the company has since abandoned can create confusion before external sources are considered.
What you do: observed behaviour
The second layer is what an organisation appears to be doing.
Recruitment activity, product pages, pricing, executive commentary and employee activity can provide evidence of its priorities and direction. Much of this material is published for operational or commercial reasons, rather than to communicate a corporate reputation.
A business describing itself as a technology leader, for example, may also have a public record of its technical work, recruitment and leadership commentary. These signals can support its stated position or raise questions about it.
They need careful interpretation. An advertised vacancy does not establish that somebody was hired. A product announcement does not prove that the product works as claimed.
AI reputation management involves understanding what these signals establish, what they merely suggest and whether they support the wider corporate narrative.
What others say: external validation
The third layer is information published or recorded by others.
News coverage, customer and employee reviews, industry publications, podcasts, forums, public filings and institutional sources can all contribute to how an organisation is described.
These sources may corroborate the organisation's claims, provide additional context or challenge its version of events. Independent evidence is particularly important because an organisation cannot establish its credibility solely through what it says about itself.
External sources are not necessarily accurate or representative. A review may describe an isolated experience. A news article may be outdated. A company directory may contain incorrect information.
Management therefore requires assessing the reliability and relevance of external evidence, not simply counting positive and negative references.
Where the three layers disagree
The most revealing findings often appear where the layers do not align.
Consider a hypothetical engineering consultancy that has repositioned itself around low-carbon building design.
Its website describes the new strategy. Its public project portfolio still consists mainly of conventional engineering work. Independent industry coverage continues to describe the business according to its former specialism.
An AI assistant asked about the consultancy might describe it primarily as a traditional engineering firm.
That description could be outdated. The company may have completed substantial new work that has not been properly documented. Alternatively, the repositioning may be ahead of the operational changes needed to support it.
The correct response depends on establishing which explanation the evidence supports. Updating old information may be appropriate in one case. In the other, the organisation needs to demonstrate that its capabilities have changed.
This is the continuing work of signal alignment, making sure what an organisation says is supported by what it does and by credible external evidence.
Adelbury's AI Reputation Audit report maps sixteen source types across the three layers, examining what each can establish and where an organisation can influence the information.
How the framework applies to leaders, individuals and brands
The same framework applies to people.
A chief executive has a declared narrative in their biography, professional profiles and public statements. Their decisions, commentary and professional activity provide evidence of what they do. Journalism, interviews, institutional records and other independent material contribute external validation.
An AI-generated account of a leader can influence how people judge the organisation they represent, particularly when the individual's credibility, conduct or expertise is important to its reputation.
For individuals, as for organisations, the task is to understand what the public record supports, correct what is inaccurate and recognise where reputational concerns arise from real behaviour.
The same layers apply to a brand: what it promises, how its products and services perform and what customers, reviewers and the media say about it.
AI reputation management vs online reputation management
Online reputation management (ORM) is an established discipline concerned with how people and organisations are represented across the internet.
It includes managing reviews, monitoring media and social discussion, improving search results, correcting inaccurate information and responding to reputational problems. Reputation management companies provide services ranging from strategic advice to review management, search engine optimisation (SEO) and lawful content removal where appropriate.
AI reputation management builds on much of this work. The difference is in how the information is encountered and interpreted.
In conventional search, users can examine individual results, visit websites and compare sources. AI tools can present information from several sources in a single answer, giving the user a ready-made account of the organisation.
That account may combine an official biography, old reporting, recent customer reviews and information retrieved from elsewhere. The person asking the question may never visit the original sources.
An organisation may therefore have a strong search presence and still receive an inaccurate or incomplete AI-generated description. Equally, it may appear frequently in AI answers without those appearances creating a favourable impression.
The disciplines overlap because they work on much of the same public evidence. AI reputation management adds the need to examine how that evidence is combined into an answer and what conclusion a reader might draw from it.
How does AI reputation management relate to GEO?
Generative engine optimisation (GEO) addresses how information is discovered, interpreted and represented in AI-led search and answer systems.
Much of the practical work concerns making content accessible, improving source credibility and understanding where organisations appear in AI-generated responses. But GEO is broader than visibility alone.
The International Association for Measurement and Evaluation of Communication's GEO Principles, published in May 2026, explicitly address accuracy, framing, omissions, source quality and reputational risk. They also distinguish visibility from outcomes such as trust and behaviour.
The question is where responsibility for reputation sits.
When reputation work is filed under GEO, it becomes part of a broader field in which much of the activity concerns technical accessibility, search performance and visibility. These are valuable activities, but reputation involves decisions that extend well beyond search.
An unfavourable AI description might originate in poor customer experience, questionable leadership decisions, misleading corporate claims or inaccurate external reporting. Resolving those issues can require operational change, legal judgement or board-level decisions.
Reputation therefore needs to be treated as a strategic responsibility in its own right, even where GEO methods and software support the work.
| Online reputation management | Generative engine optimisation (GEO) | AI reputation management | |
|---|---|---|---|
| Main question | What can people find about us online? | How is information about us discovered, cited and used in AI answers? | Does the picture AI gives of us align with what we say, what we do and what others say about us? |
| Typical work | Reviews, search results, media monitoring and corrections | Content accessibility, source credibility, technical optimisation and tracking appearances | Aligning what we say and do, online and offline, with what others say about us, then reviewing AI answers over time |
| Who usually leads | Communications, marketing or a reputation management firm | SEO, digital or marketing teams | Communications, coordinating across the organisation |
How to manage reputation in AI
AI reputation management begins with understanding the current picture and continues through investigation, action and reassessment.
A useful process has four continuing responsibilities: establish what AI says, identify the evidence behind it, decide what needs to change and check whether the picture changes over time.
Start with an AI reputation audit
The first step is to establish a baseline.
An AI reputation audit tests how different AI systems describe an organisation, brand or individual. It should cover the questions that actual stakeholders might ask, including questions the subject would prefer not to receive.
For a company, these might concern its credibility, leadership, financial position, workplace practices, customer experience or comparisons with competitors.
The audit records the answers across selected platforms, identifies recurring descriptions and examines the evidence associated with them. Findings are then assessed against the three layers.
It should distinguish a pattern supported by several observations from a single unusual answer.
An audit establishes what was observed at a particular time. Management takes those findings forward, assigns responsibility for addressing material problems and continues checking the public record.
Diagnose the problem before deciding what to change
An unfavourable AI answer does not automatically indicate a reputation problem of the same kind in every case.
Four situations require different responses.
The AI-generated answer is wrong. A system might invent an allegation, confuse two people with similar names or attribute a statement to someone who never made it. There may be no underlying source to correct. The task is to document the error, check how widely it recurs and make accurate evidence available through appropriate channels.
The information is outdated or incomplete. An AI answer may reflect an old leadership team, an abandoned product or a former company name. The public record may need updating, or current information may be difficult for the system to retrieve.
The public record contradicts itself. Official claims may be inconsistent with operational evidence or independent reporting. The contradiction requires investigation because neither the organisation's preferred account nor the external material can automatically be assumed correct.
The criticism is accurate. An AI system may fairly summarise genuine customer complaints, operational failures or documented leadership decisions. Changing the wording on a website will not resolve the underlying issue. Management may require the organisation to improve its conduct and demonstrate the change.
| Situation | What it looks like | Typical response |
|---|---|---|
| The AI answer is wrong | An invented allegation, two people confused or a statement wrongly attributed | Document the error, check how often it recurs and make accurate evidence available |
| The information is outdated or incomplete | An old leadership team, an abandoned product or a former company name | Update the public record and make current information easier to find |
| The public record contradicts itself | Official claims at odds with operational evidence or independent reporting | Establish which account the evidence supports before acting |
| The criticism is accurate | Genuine complaints, operational failures or documented decisions, fairly summarised | Address the underlying issue and show the change. New wording will not fix it |
A further risk is deliberate manipulation of the public record through fabricated allegations, impersonation or coordinated publishing. Adelbury's article on the changing economics of reputational attacks examines how AI can reduce the cost of producing such material.
These cases still require evidence-based investigation. Legitimate criticism must remain distinct from fabricated or deceptive activity.
Improve the evidence and address the underlying problem
Once a problem has been diagnosed, the response should follow the evidence.
For declared narrative, that may involve updating corporate information, clarifying a company's identity, correcting executive biographies or providing verifiable support for important claims.
For observed behaviour, it may mean documenting completed work, ensuring products and services are represented accurately or addressing operational failures.
For external validation, the response may involve requesting factual corrections, providing information to journalists, maintaining accurate public records or earning independent recognition through substantive work.
Some problems can be resolved through communications. Others require action by different functions within the organisation. The task is to identify who can address the cause, not simply who can produce more content about it.
AI reputation optimisation (AIRO) is the work of making the public evidence about an organisation accurate, consistent and easy for AI to find and interpret, so the picture AI gives of it matches what is true. Management is the wider task, which also includes diagnosis, organisational decisions and continuing assessment. Neither gives control over the answers produced.
Review the picture over time
AI answers can change as models are updated, new information appears and different questions are asked.
Research published by SparkToro in January 2026, conducted with Patrick O'Donnell of AI-monitoring company Gumshoe.ai, collected 2,961 responses to twelve brand and product recommendation prompts across ChatGPT, Claude and Google's AI tools. The researchers found substantial variation in the lists and ordering of recommendations.
The study examined recommendations rather than the factual accuracy of reputational descriptions. Its findings nevertheless illustrate why a single answer is a poor basis for measurement. The research also involved companies with a commercial interest in the market.
Ongoing monitoring should therefore use a defined set of stakeholder questions, documented testing conditions and repeated observations.
The objective is to identify material changes and persistent problems, not to treat every variation in wording or position as a reputational event.
Where AI reputation management tools fit
AI reputation management software can support monitoring by gathering responses, tracking mentions, identifying recurring themes and comparing results across platforms.
Some products also provide citation tracking, sentiment analysis and competitive comparisons. These can be useful inputs, especially when monitoring many questions or organisations.
But a tool cannot establish whether an unfavourable description is deserved merely by assigning it a negative sentiment score.
Interpretation requires examining the evidence, understanding the organisation's circumstances and deciding what response is appropriate.
The distinction matters when choosing between reputation management tools, software and consultancy services. Monitoring produces observations. Reputation management uses those observations to inform decisions and action.
Who is responsible for AI reputation?
Communications should coordinate AI reputation management because its responsibility extends across the organisation's public reputation.
But the evidence AI systems may encounter sits across several functions.
Marketing is responsible for much of the brand narrative and product positioning. HR has responsibility for recruitment information and employment practices. Investor relations manages financial communications. Legal teams address disputed claims and regulatory issues. Product and operational teams are responsible for what the organisation actually delivers.
Leadership and boards may need to act where reputational findings reveal material business risks, contradictions in strategy or problems requiring organisational change.
Communications cannot control all these activities. It can maintain a view across them, identify inconsistencies, bring the relevant people together and ensure that reputational consequences are considered.
This is also the argument Adelbury's founder, Stephen Davies, makes in his PRMoment article, *When AI becomes your reputation analyst*.
The coordinating responsibility becomes especially important when different teams are unaware that their publicly available information presents contradictory accounts of the organisation.
What can and cannot be managed?
AI reputation management gives organisations ways to improve their public evidence and respond to reputational problems. It does not provide direct control over AI-generated answers.
What an organisation can influence
Organisations can maintain accurate information across their own websites, profiles and publications. They can make verifiable facts easier to discover, provide supporting material for significant claims and seek corrections when external sources contain errors.
They can also improve the behaviour that creates reputational problems, document the resulting changes and build credibility through independent evidence.
These actions have value beyond AI. Accurate information and credible behaviour also support conventional reputation management, stakeholder relationships and public trust.
They provide AI systems with better material to draw on, although they cannot determine how that material will be used.
What an organisation cannot control
An organisation cannot guarantee how an AI system will describe it, which sources it will retrieve or cite and whether a correction will change future answers.
Different products and prompts can produce different accounts. Some answers draw on older information represented within a model, while others use current retrieved sources. The information available and the system's interpretation may change without notice.
Monitoring also has limits. A set of tested answers cannot establish what every user sees or prove that a reputational change has caused a particular commercial outcome.
There is a further distinction between making information accurate and trying to make every answer favourable. Independent criticism, where properly supported, is part of the public record. Reputation management cannot legitimately require its disappearance.
A useful assessment therefore considers whether descriptions are accurate, appropriately supported and materially relevant, while recognising uncertainty about their reach and influence.
Where should you begin?
The starting point is to establish your AIR: how AI currently describes you and whether the available evidence supports that picture.
Adelbury's AI Reputation Audit report sets out the three-layer framework and a sixteen-source matrix for assessing the public record. It provides a method for identifying inconsistencies, tracing evidence and prioritising action.
An initial audit can establish the baseline. Ongoing management then involves addressing the findings, assigning responsibility and reassessing the evidence as circumstances change.
Adelbury runs AI reputation audits for organisations that want to set that baseline. To discuss continuing reputation management, contact Adelbury.
Talk to Adelbury
