Skip to content

How do you use AI, and is our data used to train AI models?

The short answer

Anthropic's Claude models write the reading of your report from your submitted answers and figures, and every AI text is labelled Written by AI. We do not train AI models on your answers. The AI never scores you, never gives a verdict and names people only by role.

Where AI is used

WhereWhat the AI does
The AI reading of your reportA short text about your results. It is written straight away when you submit, in English, and takes about half a minute.
The team readingWritten once when everyone on the team has finished.
Claims and evidenceWhere you answered Yes to a claim, the AI reading checks whether your written answer backs it up; it does not change your score.
Where your written answers differOnce everyone has finished, AI compares your written answers to the same questions and shows up to five where they differ.
The free AI mentorElevenLabs runs the voice call, Anthropic prepares the call and, if you continue later, notes where you stopped.

What the AI does not do

  • Your score is computed by rule from the answers; the AI reading does not change it.
  • Every figure in the reading comes from your answers or from the cards in Your numbers.
  • The readings and the mentor give you advice to consider; they do not make decisions about you.
  • To report a mistake, select Something wrong in this reading? under the text and say what is wrong in your own words.

Training on your data

We do not train AI models on your answers. Anthropic's Claude API writes the readings, and Anthropic's commercial terms say it may not train models on customer content from its services.

If you connect your own AI tool under Settings > AI, what it reads goes to the company that provides it, under that company's own terms and privacy policy. Some AI plans for individuals may use what you share to improve their models unless you switch that off in the tool's settings.

Where this comes from

Read on 31 October 2026.