About
Mission
The Correlation Index is an independent research project that measures, with public evidence, how concentrated financial institutions are on a handful of AI foundation models β and the degree of empirical correlation between those models on simulated risk decisions.
The problem
As critical financial decisions β credit, underwriting, trading, fraud detection β get delegated to AI agents, those agents actually rely on a very small number of shared foundation models. Institutions that look diversified and competitive can therefore fail in a correlated way when the same shock hits β exactly like quant funds in August 2007. This risk is already documented by the Bank of England, Gallagher Re, and Stanford researchers (see the full references), but no public tool measures it today, continuously and with sources.
Who this project is for
Two audiences, at this stage. First, journalists, researchers, and risk analysts, who need to be able to cite our figures and methodology. Second, reinsurers and insurers specializing in AI risk, who need a serious, sourced signal on the concentration of this systemic risk β not a sales pitch.
What we publish
A map of public relationships between financial institutions and foundation model vendors (data explorer), and an empirical correlation test between models on risk-decision scenarios (correlation test). The method's detail, its confidence levels, and its limitations are on the Methodology page.
The team
This project is run by a small independent research team. We deliberately stay sober on personal information here β this project's credibility rests on its data and its method, not on who's behind it. For any methodology, press, or partnership question, check the Methodology page first.