Traditionally, reputation management has begun with a familiar question: What do people believe about us? This process helped to identify inaccurate information, track emerging issues and intervene before small problems become bigger ones.
The paradigm has changed so that now it will need to include two other questions.
- What does AI say about us?
- What do we train AI to say to our consumers?
Monitoring external AI systems effectively will only be the beginning of the challenge because companies are integrating AI into the consumer experience.
For communications teams, this creates a very practical question: who decides how that agent should speak?
Pelin Ateşkaya of OptimoreGroup envisions that existing communications strategy will make a dramatic shift. She notes that brands already have made extensive investments message frameworks, spokesperson preparation, Q&As, tone of voice and crisis protocols.
“An AI agent will need the same guidance, but it will be having thousands of different conversations, responding in real time and adapting its answers to individual users,” Ateşkaya explains. “This means communication strategies will need to move beyond message architecture and start defining conversation architecture: what the brand can say, what it should not say, which sources it should rely on, when it should escalate to a human and how it should respond when challenged.”
Whether it is training an internal system or monitoring external AI tools, it means there is a central question: What do the machines believe about us?
Tina Charpentier of Padilla believes that question has become important and will take a central position in reputation management.
“AI-generated answers are becoming an increasingly important layer between organizations and their stakeholders,” Charpentier explains. “Those answers can reinforce a reputation — but they can also perpetuate outdated information, amplify misinformation or create an incomplete picture of an organization.”
And unlike traditional search, an AI platform may present that conclusion as a direct answer. In 2026, companies started the process to discover what ChatGPT and other AI systems say about an organization. That monitoring will become as routine part of reputation management as monitoring press coverage.
When an AI answer becomes a belief
In all of the response of AI, accurate information Roger Hurni of Off Madison Ave puts the reputational risk simply: “A wrong AI answer about your brand becomes your customer’s belief.”
His concern stems from the same behavioral change affecting AI discovery. If an AI assistant incorrectly characterizes a company’s product, history, leadership or reputation, the user may never know there was a discrepancy.
Nothing on the page gives it away, but false placeholders are swapped out or non-related sites. That metrics moved from 0.17% results in September to 8.0% in October 2026.
Previously, users experienced a company one point at a time. They could read articles, listen to podcasts, so some online searches to help formulate an answer. Many times, they would only do one of these at a time.
AI is trained to look for the answer to the question that you ask. AI also can synthesize all of these sources to formulate an answer and to paint a portrait based on the perspective provided by the input.
The current state of many AI engines is that they make live searches on tools like Bing and Google for information that does not exist in trained LLM data. The stakes for this issue are even greater with recent research from Nozzle.io pointing to Google poisoning SERPs. This issue is that Google serves complete, normal-looking search engine results pages (SERP) whose results have been swapped out.
The experience and that an AI tool “is directly telling consumers what it thinks is the correct answer about what your company is, how it operates and if it’s trustworthy,” Jake Neerland of Padilla explains. “That summary is being treated as truth and that’s where we’re seeing those reputation problems.”
With source issues and SERP poisoning Organizations that ignored search visibility could lose traffic and eventually business to competitors. But the AI environment changes the nature of the output. A company buried on the second page of search results still existed in the results.
The technology is no longer just directing people toward information. It is interpreting the information for them. That makes AI an increasingly important reputation surface.
An AI assistant can instead tell a user what it believes the company is, whether it is trustworthy or which competitor is a better option. If the opinion of AI is that your company is not trustworthy, it can simply not be included in indirect searches.
AI reputation audits are necessity
This year has been filled with companies that are looking to see what AI tools say about them and their brands. That process is one that will become a routine part of business due to the speed and endless sources of information that AI devours into its databases.
Stefan Pollack of The Pollack Group expects a new question to appear regularly on CEO and communications agendas: “What do ChatGPT, Gemini and Perplexity say about us?”
That means going beyond typing the company name into a chatbot once. Organizations will need to test multiple platforms and multiple questions. They should understand which attributes consistently appear, whether descriptions are current, what sources influence the answers and where persistent gaps or errors exist.
With the translation capabilities, companies should be looking across languages and markets as well. Johannes Martschin of Martschin & Partner expects these audits to become as commonplace in Austria, Germany and Switzerland as conventional media roundups.
Elizabeth Marshall Black of K Harvey Brand Partners goes further, predicting that AI reputation reviews will become part of standard agency retainers and that AI misinformation will eventually earn a dedicated place in crisis plans.
Jackie Liu of The Pollack Group offers a useful counterpoint to the idea that this requires communications professionals to reinvent everything they do.
The platforms may be new, she argues, but the work is manageable. Communications teams have always researched sources of influence, monitored reputation and strengthened credible third-party information. ChatGPT, Gemini and other platforms can become another layer of that existing discipline.
Complications with local context in machine reality
International organizations face an additional challenge: AI does not always respect geographic boundaries as neatly as communications teams do. Martschin sees this in the DACH region, where AI systems can combine information from Germany, Austria and international sources despite meaningful differences in regulation, market structure and cultural context. A statement accurate for Germany may not automatically be accurate in Austria.
The risk becomes greater in fields such as healthcare and science, where an incorrect summary involving treatment, regulatory approvals or research can carry substantially higher stakes than an ordinary brand-description error.
Imma Folch of LF Channel argues that this is also where human interpretation becomes essential. AI can flag that something is being said. It cannot always determine whether that information matters in a particular market, why it is gaining attention or whether it represents noise or the beginning of a serious issue.
That requires people who understand local media, institutions, culture and stakeholders.
As monitoring becomes increasingly automated, the value of communications counsel moves toward interpreting the signal and deciding what action to take.
Ways to start reputation monitoring
What should that discipline look like? Hurn makes a key point that brand and reputation teams should keep in mind: “You have no control over the AI answer. You do have control over what the AI reads.”
When finding the right prompts to track it is important to ask what a customer wants, not was a communications team will ask. For example, a customer may never ask, “What is Company X’s reputation?”
Instead, they may ask which company is safest, most reliable, appropriate for a particular need or best suited for a specific situation. Those questions can reveal very different reputational narratives.
Fix the source, not just the symptom
AI answers change as sources change, models change and new information enters the ecosystem. That changes how communications teams think about correction. When a newspaper publishes an error, a company can contact the publication. When a website contains incorrect information, the organization can often identify the page and pursue a correction.
AI-generated misinformation can be harder to trace. The answer may represent a synthesis of information collected from multiple places.
Martschin notes that there is no single editorial office to call. Instead, organizations may have to work backward through the information environment: owned content, structured data, third-party profiles, authoritative databases, scholarly information or other sources AI systems may be interpreting.
Charpentier makes a similar argument. Managing AI reputation will be less about “fixing the chatbot” than ensuring credible, consistent information appears in sources that AI systems trust.
That requires cross-functional thinking.
Kayleigh Hanlin of True Digital Communications argues that AI findings should not live in a separate monitoring report. They should feed back into the overall marketing and communications strategy.
If AI repeatedly misunderstands a company’s services, that may expose a website problem. If it cannot identify the organization’s expertise, that may reveal a content or earned-media gap. It might also be an indicator of a brand evolution that is being muddled by outdated references and content.
If third-party sources tell a different story from the brand’s own channels, there may be a broader reputation problem. The AI answer becomes a diagnostic tool.
Earned media becomes part of reputation infrastructure
Liu expects communications professionals to continue identifying the sources that influence public understanding. This also makes earned media increasingly relevant to machine reputation.
The logical proof for the value of earned media is straightforward.
AI systems encounter what organizations say about themselves. They can also encounter what journalists, analysts, experts, social channels and other independent sources say about them.
When those sources reinforce one another, the organization’s identity and authority become clearer. When they conflict, the model has to reconcile the discrepancy.
That makes reputation less about a collection of individual media placements and more about the cumulative information ecosystem surrounding an organization.
The goal is not to manufacture a perfectly controlled narrative. That isn’t realistic with human audiences and it is even less realistic with AI. The objective is to make accurate information more available, more consistent and better supported.
Crisis Planning Needs a Machine Layer
The coordinating practice with reputation management is crisis preparedness. The same logic should will need to be adopted in crisis preparedness.
Imagine an organization facing a fast-moving issue in 2027. The communications team monitors traditional media and social conversation. But customers, employees, investors and reporters are also asking AI systems what happened. Companies need to understand the parameters they will be facing in a crisis including questions like:
- What are those systems saying?
- Are they repeating outdated facts?
- Are they confusing allegations with confirmed information?
- Are they surfacing an old issue because the available source material is stronger than new reporting?
- Are they drawing information from a different market with a different regulatory framework?
Teams may need pre-established procedures for testing AI platforms, identifying source problems and rapidly publishing clear, authoritative information that helps correct the broader information environment.
That does not give the organization control over an AI system. It gives the AI system better evidence to work with. This shift also proved the prediction from Ateşkaya at the opening of the article. As companies adopt the use of AI to manage customers and inquires, crisis management responses will need to be part of that management strategy.
It’s Reputation Management, Not AI Control
There is a temptation whenever a new communications channel emerges to imply it can be “managed” in the same way as an owned website, traditional media or advertising campaign.
AI does not work that way. Models change. Sources change. Answers can differ by prompt, user, product, geography and moment. Organizations should therefore be cautious about promises that GEO or AI reputation management can guarantee a specific answer.
What communications teams can do is much more practical:
- Improve factual consistency.
- Publish useful and authoritative information.
- Build credible third-party validation.
- Monitor recurring inaccuracies.
- Identify the information sources influencing them.
- Measure whether the narrative changes over time.
That is reputation management—not control.
Human and machine reputation are becoming one system
Ultimately, “reputation for humans” and “reputation for machines” may be a temporary distinction.
The two increasingly feed one another. People have experiences with organizations. Those experience surface as content in reviews, social conversations, media coverage and public information data. AI systems synthesize that record. Their answers then influence what another person believes about the organization.
The cycle begins again.
That is why it matters when Charpentier asks, “What does AI believe about us?”.
It is not because machines have opinions in the human sense. It is because their synthesis increasingly shapes human perception.
By 2027, the organizations best positioned to protect their reputations will not be those trying to control every AI answer. That is neither realistic nor sustainable.
They will be the organizations that systematically build an information environment in which the accurate answer is the easiest one—for people and machines—to find.
Other Partner Insights on Reputation Management
- Artificial intelligence and communication: how we use it at LF Channel
- AI and Sustainability Reporting: What Companies Need to Know
- Before the crisis
- Five tips for protecting your reputation online
- How PR & GEO Can Boost Travel Destination Visibility in AI Search
- How to prioritize GEO in your communications efforts.
- Improving your online reputation management
- The EU AI Act’s transparency rules and editorial responsibility in PR practice
- Why Most Brand Messaging Fails Before the Launch
- Why Your Brand’s Words Are Its Most Underrated Asset
