Poster and research stream on using grantmaking data to surface and govern AI-related federal spending. The work germinated at CDT after limited traction on the OMB memo guidance; I pushed the idea through iterative public artifacts to prove value and build momentum.

The phrase "AI-related discretionary spending" appears to have no search results before I used it publicly; it later shows up in a federal AI action plan (cannot prove causality, but the timing is notable).

Early framework to score and track accumulating "transparency debt" across deployed AI systems - balancing usability, governance, and auditability.