AI hallucinations emerge as a new reputational risk for businesses: Ambika Sharma

MUMBAI: AI is rapidly becoming a new source of information about brands, but its answers are not always accurate. From fabricated citations and incorrect company information to AI-generated claims that can damage reputations, hallucinations are emerging as a growing business risk. As more consumers turn to AI tools for recommendations, comparisons and research, companies can no longer afford to treat these inaccuracies as isolated errors. In this guest column, Ambika Sharma, product architect, NeuroRank, and chief strategist, Pulp Strategy, examines how AI hallucinations can reshape brand perception, the emerging legal risks for businesses and why companies need to actively monitor how they are represented across AI-generated responses.

There is an old saying that goes, “A lie can travel halfway around the world while the truth is putting on its shoes.”

In the age of AI, not only do falsehoods travel faster, they are also being manufactured by machines. These AI-generated inaccuracies, known as hallucinations, are emerging as a new reputational risk for businesses. Yet many companies are still underestimating the damage they can cause. A single false claim can shape how customers and investors view a brand, hurting trust long before the truth catches up.

*For instance, in May this year EY (previously Ernst & Young) withdrew a 44-page report after it was found to contain AI-generated hallucinations, including fabricated data, fake citations and references to research that did not exist. While the report was eventually removed, the episode underscored how AI-generated inaccuracies can undermine the credibility of even the most established professional services firms. It is still early days for AI, yet incidents involving AI hallucinations are already making headlines. More importantly, it has started affecting how people perceive brands. This example served as a reminder of the reputational risks AI-generated inaccuracies can create.*

However, companies are increasingly challenging AI companies in court over inaccurate AI-generated responses. Just as recently as May this year, a court in Munich ordered Google to stop its AI Overviews from making false claims about two German publishers after the AI incorrectly portrayed them as running scams and subscription traps. Google argued that it was merely surfacing third-party content. The court disagreed, ruling that because the AI generates and synthesises its own responses, Google is responsible for what it says.

The Munich court’s ruling highlights a much bigger issue. Every day, AI tools such as ChatGPT, Gemini, Claude and Perplexity answer questions about businesses. They recommend products, compare services and summarise company information for millions of users. When those answers are inaccurate, the consequences extend well beyond a factual mistake.

This is where NeuroRank’s ORHL framework comes in. ORHL stands for Omitted, Replaced, Hallucinated and Zero Leads, which are the four common ways AI models can misrepresent a business. A company may be omitted of a recommendation where it is the obvious choice, replaced by a competitor, described with incorrect information, or fail to appear in AI-generated responses altogether. Identifying these issues is the first step towards correcting them.

The problem has not been limited to one company or one country. In 2024, Canada’s Civil Resolution Tribunal held Air Canada liable after its chatbot invented a bereavement refund policy that did not exist. A year later, Deloitte partially refunded the Australian government after a report prepared for a federal department was found to contain fabricated quotes and citations. Together, these incidents show that businesses are increasingly being held responsible for what their AI systems say and produce.

For most companies, however, the greater risk is unlikely to be a lawsuit. It is the thousands of everyday conversations AI has with potential customers. Someone asks an AI assistant to compare products, explain a company’s pricing or recommend a service provider. If the response is inaccurate, the customer may never visit the company’s website or verify the information. Instead, the AI’s answer becomes the basis for a decision.

Three things make this particularly challenging for businesses.

First, AI hallucinations are not isolated incidents. As AI tools are used more widely, inaccurate or fabricated responses continue to surface despite improvements in the technology.

Second, people increasingly rely on AI-generated summaries instead of visiting multiple websites. Most users accept the answer at face value and move on without checking the underlying sources.

Third, the commercial impact often arrives long before any legal consequences. An inaccurate AI response may never make headlines or reach a courtroom, but it can quietly influence customer decisions, reduce trust and divert business to competitors.

The impact can be understood across three broad levels of risk. The most common are factual errors, such as incorrect product details, outdated pricing or missing information. While individually small, these mistakes can gradually affect customer acquisition and sales.

The next level is reputational. AI may incorrectly summarise customer sentiment, misrepresent a company’s products or repeat inaccurate claims as fact. Such errors can shape how customers, investors and business partners perceive a brand.

The most serious, though less frequent, cases involve false allegations of wrongdoing, fabricated compliance failures or incorrect links to scams or controversies. These can expose businesses to legal disputes, regulatory scrutiny and significant reputational damage.

Businesses have long monitored their search rankings, public messaging and online reputation. As AI becomes a primary source of information, they also need visibility into how these systems describe their brands. Monitoring AI-generated responses, identifying recurring inaccuracies and correcting them at the source will become an increasingly important part of brand management.

The law will continue to evolve as AI becomes more deeply embedded in everyday life. But businesses should not wait for a courtroom battle to discover what AI is saying about them. They should address AI inaccuracies at the source before they shape customer perceptions. After all, when AI becomes a storyteller for your brand, ensuring it gets the facts right at the very outset becomes part of protecting your reputation.

Note: The views expressed in this article are solely the author’s and do not necessarily reflect our own.