This audit measures whether your investment decisions, risk assessments, and regulatory judgements remain yours or have shifted to AI models. It examines where you still think independently and where algorithms now drive your conclusions.
When regulators ask 'why did you make this decision', your answer must stand without mentioning any AI tool. If it cannot, the decision is not yours.
Track which analysts in your organisation consistently spot risks before the models do. Protect their time and autonomy. They are detecting failure modes the algorithms cannot see.
Create a live spreadsheet where you compare decisions made with the model versus decisions made against it over the last year. If one side shows consistently better risk-adjusted returns, you are over-trusting that source.
Run a fire-drill: choose a portfolio or book from last month and rebuild the investment thesis from raw data without touching your AI tools. If this takes more than twice as long as the original decision, your analysts may no longer know how to think independently.
Ask your compliance team: if we had to defend this decision to a court, could the judge understand it from our documentation alone, or would we need the vendor to explain what their model meant?