Governance & law

How to report Deep Narrative Intelligence to your board

October 24, 2026 6 min read
Kamel Malek Director of SetRecord.ai

Board members do not need to understand transformer architecture. They need to understand risk, exposure, trend, and action. This article translates Deep Narrative Intelligence into the language of board reporting — the three metrics that hold up in a board pack, the cadence that keeps attention without overwhelming it, and the framing that connects AI narrative risk to the categories boards already govern.

Why boards should care

A board that oversees reputation risk, regulatory compliance, and cyber resilience is already overseeing the domains that AI narrative distortion touches. A false claim in a model's answer about your company is a reputational risk (it shapes what stakeholders believe), a regulatory risk (if it touches financial disclosure or ESG claims), and an operational risk (if it misleads customers or partners). The board does not need a new committee; it needs a new data feed into its existing risk oversight. DNI provides that feed.

The three metrics that drive decisions

Of the five signals a DNI programme tracks, three belong in a board pack. The first is the AI Image Score: a composite measure of accuracy, presence, and sentiment across the model panel, expressed on a 0–100 scale. Boards understand a score; they understand a trend line; they understand a target. The second is the critical-claim error rate: the share of factual claims on board-relevant topics (financials, leadership, regulatory status, safety) that are verifiably wrong. This is the metric that maps directly to disclosure obligations. The third is correction persistence: the percentage of corrected claims that remain corrected after 90 days. This tells the board whether the programme is working, not just running.

Don't know what AI says about you today?

A SetRecord audit queries supported AI models about your organisation and measures the gap with verifiable facts — within five business days.

Request an audit

The reporting cadence

Quarterly is the right cadence for most boards. A DNI quarter pack contains: the current AI Image Score with a 12-month trend, the critical-claim error rate with the top five errors flagged, the correction persistence rate with a note on any regressions, and a one-page narrative summary in plain language. Monthly reporting is appropriate for boards in sensitive periods — pre-IPO, during a crisis, during a major M&A transaction. Weekly reporting is almost never appropriate at board level; it belongs with the operational team. The discipline is to report regularly enough that the board builds intuition for the metric, but not so often that it becomes noise.

Framing it as risk, not novelty

The biggest mistake in board reporting on DNI is framing it as a novel technology topic. Boards tune out novelty. Frame it as risk oversight: "We measure what AI models say about us, the way we measure what the press says about us, the way we measure what our cyber posture looks like. Here is the current reading, here is the trend, here is what we are doing about it." This framing places DNI in a category the board already knows how to govern. SetRecord structures its board-level reporting around exactly this framing, because a metric the board does not understand is a metric that does not get resourced.

What not to say

Do not report model-by-model variance at board level; it is noise. Do not report every minor factual error; report the ones that touch board-relevant topics. Do not promise that a correction will hold permanently; report the persistence rate honestly, including regressions. Do not present DNI as a solution to disinformation in general; present it as a measurement and correction programme for your organisation's specific AI narrative. Precision builds trust; over-promising destroys it. The board needs to trust the metric before it will trust the programme.

See what AI says about you

A SetRecord audit checks what supported AI models say against your verifiable facts — within five business days.