# AI company diligence

Eight checks for diligence on an early-stage company that depends on a model: data, supplier, security, evidence, claims, rules and ownership.

Canonical URL: https://worldenterprise.org/briefs/ai-company-diligence/
Last updated: 2026-10-11
Publisher: World Enterprise, run by World Enterprise Group Incorporated

Investors who look at an early-stage software company are used to checking the entity, the cap table, the customers and the code. A product that depends on a model adds questions those checks do not reach, and this brief is organized around them.

It sets out eight checks, running from the model's role and the data behind it to the supplier, security, proof of performance, public claims, regulation and ownership of outputs. Each check names the public sources behind it, describes the documents a company might hold, and says what that evidence cannot establish.

The sources come from standards bodies, regulators, a security project and model suppliers' own documentation. The brief reports what they say and keeps its analysis apart, marked as analysis.

It makes no assessment of any company, offers no recommendation to invest, and names no company. It closes with questions to put to a company or to counsel.

This brief is general information for investors. It is not legal, tax or investment advice.

## What is inside

- The question this brief answers
- About the evidence
- How to read the checks
- Check one: what the product asks of the model
- Check two: who may use the data
- Check three: the supplier and the license
- Check four: how the product is attacked
- Check five: evidence that the product works
- Check six: what the company says about its AI
- Check seven: the rules that may reach the product
- Check eight: who owns what the product produces
- The checks side by side
- Questions to ask before relying on the answers
- What this brief cannot tell you
- Sources