AI-Written Pages in the Board Pack: Label What the Machine Drafted

Somewhere in the last board pack, a page was drafted by an AI tool. A summary, a first pass at a risk description, a tidied-up metrics narrative. Nobody meant to hide it. It simply never occurred to anyone to say so.

Directors make decisions on those pages. They deserve to know which parts a machine drafted and, more importantly, who checked them. This is not a ban. It is a labeling habit, and it costs almost nothing.

Key takeaways

• AI-drafted content is already in board packs; the question is whether anyone says so.

• Directors are owed two facts: what was machine-drafted and who verified it.

• The label protects the writer as much as the reader.

• Verification means a named person read it against the source, not that someone skimmed it.

• A simple footer convention keeps the habit cheap enough to survive.

Step 1: Adopt the two-line label

The convention can be two lines in the footer of any board paper: "Drafted with AI assistance. Verified by [name], [date]." Everything a director needs is in those two lines: the provenance and the accountable human.

The label is not an apology. It is the same standard applied to any source a paper draws on, and it lets directors weigh the page accordingly.

Step 2: Define what verification means

Verification is the load-bearing word. It means a named person read the draft against the underlying data and documents and takes responsibility for what it says. It does not mean the draft was skimmed for tone.

For the pages that carry numbers or claims about risk, verification should include a check against the source figures. A polished wrong number is worse than an ugly right one, because it invites confidence it has not earned.

Step 3: Decide which papers allow drafting help at all

Some boards will be comfortable with AI-drafted summaries of public material and uncomfortable with machine-drafted risk assessments. That is a legitimate line to draw, and drawing it in advance beats discovering the disagreement after a director asks.

This fictional example shows one house rule. It is an illustration, not a prescription.

The two-page cyber briefing covers the format the label rides on, and the questions before approving an AI initiative apply to the tools doing the drafting.

Step 4: Make the habit cheap enough to keep

Governance habits die of friction. Two footer lines, one named verifier, one house rule about which papers qualify. Anything heavier will be abandoned by the second board cycle, and an abandoned convention is worse than none because it teaches people the labels are decorative.

Frequently asked questions

Is AI-drafted material in board papers actually a problem?

Only when it is invisible. Drafting help is fine; directors weighing a page without knowing its provenance is not. The label converts an awkward discovery into a stated practice.

Won't labeling make directors distrust good work?

Experience runs the other way. Named verification builds trust because the accountability is explicit. What erodes trust is discovering later that a confident page was never checked.

Who owns the convention?

The corporate secretary, with the chair's backing. They control the pack, so they can enforce the footer the way they enforce every other format rule.

Does this cover confidential data going into AI tools?

That is the separate, harder rule: what information may leave the building at all. Your AI policy should already answer it; the labeling habit assumes that rule is being followed.

A Defensible Next Step

At the next pack deadline, add the two-line footer convention and name the verifier for every paper. Book a Decision-Clarity Call if you want help drafting the house rule.