About
Every action creates liability. The economy has always answered it the same way: quantify it, price it, insure it.
AI now recommends, decides, and executes inside production systems. Its liability accumulates unwritten, because it is unmeasured.
North* grades the liability surface of AI in software: a source-anchored registry of every role AI holds, resolved to a reproducible grade.
The standard
Quantifying AI's liability takes a standard: a role taxonomy, permission enums, a materiality model, a report schema — open, and verifiable against source.
Not “AI-powered.” Not “uses machine learning.”
Each role, pinned to repository, path, function, and commit. Two reviewers, same code, same grade.
Underwriters price from the grade. Buyers diligence with it. Operators disclose through it.
[enter] Research