Institutional investors are already using AI to review disclosures before they make investment decisions. A May 2026 survey of institutional investors by the Center for Audit Quality and KRC Research, identified that AI is used on company filings to extract financial or operational metrics (60%), summarize key sections (58%), and to compare peer-company disclosures (50%). Further ways AI is being used include to identify inconsistencies between narrative and numbers, compare disclosures across periods, synthesize insights across multiple filings, and to flag anomalies.
Regulation is moving in the same direction. The joint data standards under the Financial Data Transparency Act took effect on October 1, 2026, introducing common identifiers, structured formats, and consistent definitions across financial regulators, including the SEC. Nothing about your filing requirements has changed yet, although further SEC rule-making will likely follow. These evolving standards fit a broader historic pattern; every disclosure regime the SEC has modernized over the last decade has moved toward machine-readable formats, and each step along that path, from scanned PDF to plain text to HTML and to XBRL, cuts the work a machine has to do before it can use what you filed.
So, a proxy statement must now be optimized for AI models, in addition to investors and proxy voting screens.
We recently held an AI legibility forum with Labrador’s Iain Poole, Managing Director, North America, and Jennifer Cooney, Managing Director, Advisory and Design. Iain and Jennifer shared a simple checklist for drafting a proxy with AI legibility in mind:
Do
- Style a real character rather than substituting a letter
- Name your columns in text
- Make alt text a label, not a file name
- Restate in text whatever a graphic asserts
- Use one term for one thing, and one name form per person
- Consider key messages and stress test for clarity
- Write footnotes that name what they qualify
Don’t
- Let a mark be an image if a character would do
- Leave a matrix without names above its columns
- Put a headline percentage only in a chart
- Change the basis of a figure without labelling it
- Alternate between an acronym and its expansion
- Assume a footnote marker carries meaning to a parser
Get a complimentary AI Legibility Review of your proxy statement
We are offering a complimentary five-method pass across your most recent proxy statement vs. the prior year.
- Legibility: what is legible and what is not under automated AI reading, covering marks, matrices, tables, and alt text
- Consistency: every figure recomputed across pay, fees, ownership, and year-over-year
- Compensation: pay mix, metrics, and outcomes recomputed
- Tone: what a machine reports you said, including how it handles nuance
- Investor Screen: whether information tied to major institutional investors’ proxy voting guidelines is AI retrievable and legible
The deliverable is a set of targeted refinements that do not change the page for a human reader.
To request a review, you can contact us below.