AI Drives Misleadingly Positive MD&A in Quarterly Reports—Securities Litigation Likely to Emerge

Blurb: A new study finds that public companies are increasingly using positive language in their quarterly MD&A reports to influence AI-driven analysis, even when financial results are poor. This trend raises concerns about misleading disclosures and the potential for future securities litigation.

In a recent study, Keren Bar-Hava, head of the accounting department at Hebrew University Business School, reveals a significant shift in how public companies craft their quarterly Management Discussion and Analysis (MD&A) reports included within their financial statements. (See here WJS Opinion article summarizing her study at: Quarterly Reports Are Written for AI – WSJ , Paywall).

Traditionally written to inform investors, these MD&A reports are now increasingly tailored to appeal to artificial intelligence (AI) algorithms. As a reaction to AI analysis, MD&A sections have become notably more positive in tone, especially when actual outcomes are bad. Companies are strategically using optimistic language to influence how AI interprets their disclosures, even when financial performance is declining.

With the advent of advanced AI tools like ChatGPT, analyzing MD&A sections for trends in tone, complexity, and word choice has become a task that can be accomplished in seconds. This technological leap has led companies to adapt their reporting strategies, knowing that their reports are being scrutinized by machines before reaching human eyes.

Bar-Hava identifies three types of AI-induced disclosure pressure influencing corporate reporting:

1. Exposure Pressure: AI tools flag vague or evasive language, pushing companies to project confidence even in uncertain times.
2. Competitive Pressure: Algorithms compare the tone of reports across peer firms, making companies strive to sound stronger than their competitors.
3. Reputational Pressure: AI-driven insights feed into analyst dashboards and investor platforms, where a poorly framed sentence can have widespread repercussions.

Bar-Hava’s study of 108 MD&A reports from 27 top U.S. firms between 2021 and 2024 found a notable trend: the use of positive language has increased, even when financial performance declined. This contrasts with earlier studies that showed a more cautious tone during periods of poor performance. Now, words like “growth,” “resilient,” and “opportunity” have become more common, while terms indicating uncertainty have diminished. Interestingly, the most positive reports often came from the worst-performing firms, suggesting a strategic use of tone to influence AI interpretations.

This shift poses a risk of creating a gap between the optimistic language used in reports and the actual financial reality. While the Securities and Exchange Commission (SEC) has made strides in encouraging clear and concise narrative disclosure, the regulation of tone remains a blind spot. As AI-driven tone scores increasingly influence market behavior, there is a growing need to treat tone as a material disclosure element.

Bar-Hava argues that boards must recognize they are writing for both human and machine audiences. Since machines interpret the text literally, the inflation of positive tone can mislead investors. To ensure truth in reporting, it is crucial to monitor not just what is said, but how it is said.

While we have not yet seen securities litigation premised on misleading tone, given the growing gap between reality and language Bar-Hava warns of, the current use of AI to influence MD&A sections may mean that such lawsuits may inevitably be forthcoming. Such a gap is exactly what plaintiffs and plaintiffs’ attorneys seek to find and correct, and they will be aided by AI tools themselves in doing so.