Rucsandra Moldovan is an associate professor of accounting at the John Molson School of Business and serves as a board member of the Canadian Accounting Standards Board (AcSB). Her research focuses on how corporations communicate with information intermediaries, investors and other participants in the capital markets.
article
When it comes to AI, is honesty the best policy?
New study examines how the stock market interprets AI-related disclosures from non-tech firms
Photo by Immo Wegmann on Unsplash
When a software company says it uses artificial intelligence, no one is surprised. When a food manufacturer, hospital, or retailer says the same thing, something more interesting is happening.
Over the past decade, non-tech firms have been adding AI-related language to their annual reports. Are investors treating these disclosures as credible information about future growth, or as noise worth ignoring?
The risk assessment of AI
A new study from Alexey Lyubimov and Rucsandra Moldovan, both associate professors of accountancy at the John Molson School of Business, set out to investigate how investors view these disclosures. With co-authors Luminita Enache (University of Calgary) and Johannes Impink (University of Florida), the team examined publicly listed non-technology companies over more than a decade to explore how AI-related disclosure in annual reports on AI relates to a firm’s cost of equity capital. This is the return investors require to hold the firm’s stock, and a direct reflection of how risky they perceive the firm to be.
The researchers extracted what companies said about AI in two parts of the annual reports: the business description, where firms tend to discuss AI as an opportunity, and the risk factors section, where firms disclose what could go wrong. They then examined how each type of disclosure related to the cost of equity over the sample period. The findings offer a more nuanced picture than the usual "AI is good/bad for valuation" narrative.
On the opportunity side, non-tech firms that discussed AI-related activities in their business description enjoyed a lower cost of equity — investors appeared to see these disclosures as informative rather than empty hype. This effect grew stronger after 2016, when two high-profile events (DeepMind's AlphaGo victory and the Sophia robot) turned AI into a mainstream topic. As public attention to AI rose, so did the informational value of what non-tech firms said about it.
The risk side is where things get interesting. Simply disclosing AI risks on its own had no clear effect. What mattered was whether a firm disclosed more or less AI risk than would be expected given its industry and characteristics. Firms that disclosed more than expected saw their cost of equity fall; firms that disclosed less than expected saw theirs rise. In other words, investors appear to read unexpectedly frank risk disclosure as a signal that management understands its exposure, and unexpectedly quiet risk disclosure as a warning sign.
Importance of addressing both sides
For investors, AI language is not a green light or a red flag on its own. Rather, the takeaway is that the informational content of AI disclosure depends on what is said and how it compares to what peers are saying. A firm that talks only about opportunities while its industry is openly wrestling with AI risks is sending a different signal than a firm that addresses both sides.
The study also points to a broader question for the corporate disclosure environment. Accounting standards were built around an economic structure that predates AI, and mandatory reporting frameworks are still catching up. For now, the conversation about AI is taking place voluntarily in annual reports and investors are paying close attention to what’s said and what isn’t.
Read the cited research, "How Does the Stock Market Interpret Non-tech Firms' AI-related Disclosures?"
