Source: Blankespoor, DeHaan & Li (2026) — “Generative AI in Financial Reporting”
The part that bothers me is how calm the language sounds. It makes the risk feel distant, when the real problem is already sitting inside today’s systems and decisions.
One of the more interesting empirical questions about generative AI is deceptively simple: are companies already using it to write their financial disclosures? This paper provides an early, careful answer: yes, measurably, and the usage is growing. The researchers examine five types of corporate disclosure — earnings press releases, conference call prepared remarks, risk factors, MD&As, and IPO filings — through 2024. Using a commercial AI detection tool validated against a known sample, they find statistically significant evidence of generative AI usage in all five types. The most striking figure: up to 4.5% of new text in certain disclosure categories in 2024 appears to have been written by AI. That sounds like a small percentage. But in the context of financial reporting, where disclosures are subject to regulatory requirements, auditor review, and investor scrutiny, any meaningful AI usage raises questions about process and quality. The paper also finds that filings with higher estimated AI usage have systematically different linguistic properties — they tend to be more fluent, more uniform in style, and exhibit patterns consistent with AI generation. Usage is higher in cross-sectional groups where you’d expect it: firms with less internal writing capacity, more boilerplate content, time pressure around filing deadlines.
In plain English, that is why the result matters beyond the chart. It changes where people should look, what they should question, and which comfortable assumption probably needs to be retired.
For leaders, the lesson is simple: if the risk timeline changes, the attention timeline has to change too. Waiting until everyone agrees it is urgent is usually how organisations arrive late.