As generative artificial intelligence (AI) continues to redefine the landscape of digital information, corporate communication professionals are facing a new challenge: ensuring that Large Language Models (LLMs) like ChatGPT, Claude, and Gemini are portraying their organizations accurately. Alex Sevigny, an associate professor at McMaster University and an adviser at Ragan’s Center for AI Strategy, warns that if communicators want to safeguard their organization’s reputation, they must proactively audit the information these models provide to the public. During a recent Ragan AI Certificate Course, Sevigny emphasized that the first step in managing an AI-driven reputation is to ask the same questions an audience would ask, identifying discrepancies before they lead to widespread misinformation.
The shift toward AI-mediated information retrieval represents a fundamental change in how stakeholders—including investors, journalists, and consumers—interact with brand data. For decades, Search Engine Optimization (SEO) was the primary tool for visibility. However, as "Generative Search" becomes the norm, the focus is shifting toward Generative Engine Optimization (GEO). This requires a deep understanding of the training data sets and the specific logic used by AI to synthesize brand narratives. Sevigny’s framework for auditing a brand’s AI presence centers on five core inquiries designed to reveal the strengths and weaknesses of a brand’s digital footprint.
The Five Critical Questions for Brand Auditing
To maintain control over a corporate narrative, communications teams must move beyond passive monitoring and engage in active testing of LLMs. Sevigny outlines five essential questions that every communications department should pose to various AI platforms.
1. The Identity Summary: "Who is this organization and what does it do?"
The most basic test is to ask an LLM to summarize the organization, its major business segments, products, and headquarters. This serves as a litmus test for the "recency" of the model’s training data. Because LLMs are trained on historical datasets with specific cutoff dates, they may rely on information that is months or even years old. If an organization has recently undergone a merger, a rebranding, or a significant pivot in business strategy, the AI may still be providing an obsolete version of the company’s identity.
2. The Competitive Landscape: "How do we compare to our peers?"
Sevigny suggests prompting the tool to "compare your organization’s competitive strengths and weaknesses" with one or two primary competitors. This question reveals how the AI perceives the brand’s market positioning. It also uncovers the "sentiment" baked into the AI’s training data. If the AI consistently highlights a competitor’s innovation while labeling the user’s brand as "traditional" or "legacy," it indicates a need for a strategic shift in the digital content being published by the brand to influence future training cycles.
3. The Leadership Ledger: "Who leads the organization and what is their background?"
Accuracy in leadership data is vital for investor relations and executive positioning. Testing an AI on the names, titles, and professional histories of the C-suite can expose "hallucinations"—a phenomenon where AI confidently asserts false information. Inaccurate leadership data can lead to confusion during high-stakes periods, such as executive transitions or quarterly earnings reports.
4. The Reputation and Sentiment Audit: "What are the most common criticisms of this brand?"
Understanding the "negative" training data is just as important as the positive. By asking the AI to summarize common criticisms or past controversies, communicators can see which historical issues the AI prioritizes. If an AI continues to highlight a resolved crisis from five years ago as a primary characteristic of the brand, the communications team must work to flood the digital ecosystem with more current, positive, and authoritative data to rebalance the model’s output.
5. The Mission and Values Check: "What does this organization stand for?"
Finally, communicators should ask the AI to describe the organization’s mission, values, and Corporate Social Responsibility (CSR) initiatives. This determines if the brand’s internal purpose is successfully translating to its external digital persona. If the AI’s description of the company’s values feels generic or misaligned with the official mission statement, it suggests a gap in the brand’s messaging consistency.
The Evolution of AI in Communications: A Brief Chronology
The urgency for AI brand auditing has accelerated rapidly over the last 24 months. Understanding the timeline of this technological shift helps contextualize why Sevigny’s recommendations have become critical for modern PR.
- November 2022: OpenAI releases ChatGPT, bringing generative AI into the mainstream and sparking immediate concerns about data accuracy and corporate reputation.
- Early 2023: Major search engines, including Google and Bing, announce the integration of generative AI into search results (SGE – Search Generative Experience). This marks the end of the "blue link" era and the beginning of summarized, AI-generated answers.
- Mid-2023: PR and communications firms begin establishing dedicated AI task forces. Organizations like Ragan’s Center for AI Strategy are formed to provide guidelines on ethical AI use and reputation management.
- Late 2023: The "hallucination" problem becomes a recognized corporate risk. High-profile cases of AI-generated misinformation lead brands to realize that they cannot leave their digital narrative to chance.
- 2024: The emergence of "Agentic AI" and more sophisticated LLMs makes it possible for AI to not only summarize info but to perform tasks. Brand auditing moves from an occasional check to a daily necessity for communications teams.
Supporting Data: The Impact of AI on Information Trust
Recent industry data underscores the importance of Sevigny’s five-question audit. According to the 2024 Edelman Trust Barometer, public trust in AI-related information remains fragile. While 63% of employees across various industries are using AI at work, only about 35% of the general public expresses full confidence in the accuracy of AI-generated corporate information.
Furthermore, a study by Gartner predicts that by 2026, traditional search engine volume will drop by 25%, with consumers turning instead to AI chatbots and other virtual agents. This shift means that the "top of the funnel" for brand discovery is no longer a website or a social media page, but an AI’s internal representation of that brand. If the AI’s internal representation is flawed, the brand loses its first and most important opportunity to make an impression.
Data from Muck Rack’s "State of AI in PR" report also shows that 64% of PR professionals are now using generative AI for content creation, but only 22% have a formal policy for auditing what AI says about their clients. This gap represents a significant reputational vulnerability that the five-question framework aims to close.
Official Responses and Strategic Implications
While tech giants like OpenAI and Google have implemented "grounding" techniques to improve the accuracy of their models, they acknowledge that LLMs are not real-time databases. In official documentation, OpenAI notes that ChatGPT is "not a source of truth" but a "statistical model of language." This distinction is vital for communicators to understand. The AI is not "looking up" the company; it is predicting what words most likely follow each other based on its training.
Industry leaders are increasingly calling for a more symbiotic relationship between PR and AI development. "We are moving into an era of ‘Algorithmic Public Relations,’" says Marcus Merrell, a digital strategist. "It is no longer enough to pitch a journalist. You have to ‘pitch’ the model by ensuring your official press releases, white papers, and executive speeches are indexed in the high-authority databases that these models crawl."
The implications of this shift are profound. If an organization fails to conduct regular AI audits, it risks "narrative drift," where the public perception of the company is shaped by a machine’s outdated training data rather than the company’s current reality.
Analysis: The Future of Generative Engine Optimization (GEO)
The transition from SEO to GEO requires a more holistic approach to corporate communications. In the SEO era, keywords and backlinks were the primary currency. In the GEO era, the currency is "authoritative context."
AI models prioritize information that is consistent across multiple high-authority sources. Therefore, the strategy for fixing an incorrect AI summary is not just about changing a website’s meta-tags. It involves a multi-channel effort:
- Updating Wikipedia and Knowledge Graphs: Since many LLMs weigh Wikipedia heavily, maintaining an accurate and well-cited page is essential.
- Strategic Press Distribution: Ensuring that major news outlets carry the latest updates about headquarters, leadership, and business segments.
- Consistency of Voice: Using the five questions to ensure that the brand’s "voice" is distinctive enough that the AI can replicate it accurately in summaries.
Alex Sevigny’s five-question framework is more than just a checklist; it is a defensive strategy for the digital age. As AI models become the primary intermediaries of human knowledge, the role of the communicator must evolve from a storyteller to a "narrative architect," ensuring that the blueprints the AI uses to build a brand’s image are accurate, current, and reflective of the organization’s true identity. Failure to ask these questions today could result in a brand legacy that is written by an algorithm that doesn’t know the full story.
