The AI Transparency Index™: A Simple Framework for all Orgs

By Karl George MBE  |  GovernAI Academy  |  old.governanceai.io

Most organisations using AI today have no common language for describing how AI contributes to their work. Ask someone whether they “used AI” for a report, and you will get answers ranging from “I ran it through Grammarly” to “I asked ChatGPT to write the whole thing.” Those are fundamentally different activities with fundamentally different risk profiles, but without a shared vocabulary, they are treated identically.

That is the problem the AI Transparency Index™ solves. It provides a six-point scale that any organisation can adopt immediately.

The six levels

AI-0: Human Only. No AI involvement at all. You wrote it, you researched it, you produced it.

AI-1: AI-Assisted. Spelling, grammar, formatting only. Grammarly, spell-checkers. The AI tidied your words; it did not change what you said.

AI-2: AI-Supported. AI helped you summarise source material, brainstorm ideas, or structure an early draft. You directed the work and authored the final content.

AI-3: AI-Partnered. AI generated substantive content that shaped the work product’s structure, argumentation, or analysis. You directed it, edited it substantially, and are accountable for the result.

AI-4: AI-Led. AI generated the majority of the content. Your role was commissioning, reviewing, validating, and refining.

AI-5: AI-Autonomous. An AI system produced this with minimal human input. Your role is oversight and exception management.

The critical principle: proportionate governance

Here is where this framework differs from most AI policies: it does not treat all AI use the same. A grammar-checked email (AI-1) needs no governance at all. A substantially AI-generated client report (AI-4) needs formal declaration, management review, and documented verification.

The framework makes this explicit through a risk matrix that combines the AI level with the impact of the work product. Low-impact AI-1 work? Do nothing. High-impact AI-4 work? Full governance. Most daily work sits in the bottom-left corner of the matrix, where no action is needed. That is deliberate. If the framework feels like bureaucracy, people will work around it, and that defeats the purpose entirely.

External transparency built in

The framework does not just govern internal use. It provides a complete external transparency approach: a website disclosure statement, an email footer that covers the firm’s general AI use, meeting disclosure protocols for AI note-takers and transcription tools, and tiered report declarations that match the AI contribution level. Templates for all of these are included in the full workbook.

Download it for free

A free summary deployment guide is available from the GovernAI Academy, giving you the complete Index, the risk matrix, template disclosures, and a deployment checklist (please fill out the expression of interest form to get yours). The full AI Transparency Framework™ Workbook, with detailed case studies, worked examples for every level, and the complete implementation guide, is available to organisations that engage with The Governance Forum through our workshops and the Governance AI Journey.

This is a tool I believe every organisation should have. The summary guide is free because governance should not be a luxury. For organisations that want to go deeper, we are here to help.

#AITransparencyIndex #AIGovernance #ResponsibleAI #GovernAI #GovernAIAcademy