Transparently AI

Posted on February 3, 2026 by Editor

An interesting piece from Vincent Huck in this week's Corporate Disclosures e-zine (registration required, but you'll thank us later) asking questions that need answers.

Companies are using AI to prepare aspects of their annual reports. Should they disclose that they do? Speaking on a webinar hosted by Falcon Windsor in partnership with Insig AI, AstraZeneca's corporate brand and communications director, Martin Blaxall, has mixed feelings. On the one hand, it's great for summarising large pieces of text. But on the other, it isn't always accurate. In one AI test, a figure that should have read '1.3m patients' became '1.3bn'. And he has plenty of other examples - good and bad.

Our view? If AI will tempt companies to take short cuts in the review processes they go through, often termed the disclosure controls and procedures, then, well, that's a problem.

On the other hand if the normal checks, cross-checks and efforts to tie down the accuracy of all sources in the report with close reviews from relevant internal and external experts, senior management, the audit committee and the board continue to work… well, perhaps that's less of an issue?

Whatever the right answer is, it should be on the minds of regulators, standards setters, the audit profession and certain policy makers, as change in this space continues to accelerate. Well worth reading the entire, thought provoking piece.