Lending & credit collections AI

Machine-learning underwriting, pricing, and credit decisioning — often fully automated, with no per-application human review, so the organizational levers all sit upstream: the choice of model, the fair-lending testing regime, the search for a less-discriminatory alternative, and the adverse-action notice that must explain a denial. Three facts shape the governance. A denial's explanation is a separately-resourced, separately-failable duty independent of statistical bias: a model can pass the disparity test and the organization can still fail by being unable to give an applicant specific, accurate reasons. Facially-neutral aggregate features — a school's default rate priced into an individual's terms — can carry protected-class impact, which is exactly what disparate-impact testing exists to catch. And the harder governance question is not whether a disparity exists but how hard the law requires an organization to search for a less-discriminatory model that performs as well — a question the record shows resolved by enforcement or left at an impasse, rather than settled. The Lab networks model only the deploying organization — its model, compliance and testing functions, and decision records; the applicants being decided sit outside the dynamics, and no credit outcome is computed on any diagram.

Use cases

What AI is doing here

ML underwriting & risk-based pricing

Predictive

Machine-learning models that approve, decline, and price credit — often fully automated with no per-application human review, so every governable lever is upstream: the model chosen, the fair-lending testing regime, and the search for a less-discriminatory alternative that performs as well.

Adverse-action reason generation

Predictive

The system that must turn a model's denial into specific, accurate principal reasons for the applicant — a separately-resourced, separately-failable duty a complex model does not discharge by being statistically accurate, and one an organization can fail even when its bias testing passes.

Fair-lending & disparate-impact testing

Predictive

The testing regime that checks a model for protected-class impact carried by facially-neutral aggregate features — and, past detection, the contested search for a less-discriminatory model, where the hard governance question is how hard the law requires an organization to look.

Case files

What has gone wrong and right

Documented deployments, presented as model organizations calibrated to the evidence, with full citations.

Automated underwriting with its fair-lending testing on the record

United States (federal — CFPB no-action letter; fair-lending monitorship via private agreement with civil-rights organizations)

Upstart's machine-learning underwriting and pricing model, which uses education and other alternative data, ran for five years under a regulator's no-action letter with a reporting duty, and the regulator published the access results: 27% more applicants approved than a traditional model, at 16% lower average APRs, with near-prime applicants approved at roughly twice the rate. It is the lending family's only regulator-verified service term — and its most detailed public fair-lending record. Four monitorship reports found no close protected-class proxies, but identified approval disparities for Black applicants, flagged a likely-viable less-discriminatory alternative, and ended in a documented impasse over how hard the law requires an organization to search. Underwriting is fully automated, so every governable lever is upstream.

Explore this deployment in the PAN Lab →

Cleared on the numbers but faulted on the explanation

United States (state financial-services supervisory investigation of a bank's card underwriting)

A bank's automated credit-decisioning for a widely used consumer card drew viral allegations of gender bias in credit-line assignment. A state regulator analyzed roughly 400,000 in-state applicants and found no unlawful discrimination on a prohibited basis — the model was cleared on the numbers. The same investigation documented failures of explanation, customer service, and perceived transparency: applicants could not learn why they received the terms they did, front-line staff could not explain the decisions, and the opacity destroyed consumer trust even though the underwriting was found lawful. The cleared-but-faulted case: a statistically clean model paired with a failed duty to explain, which regulators treat as a separate obligation a black box does not discharge by being accurate.

Explore this deployment in the PAN Lab →

The governance an enforcement action had to write

United States (state attorney general fair-lending enforcement; Assurance of Discontinuance)

A state attorney general reached a $2.5 million settlement with the student-loan lender Earnest over its AI underwriting — the domain's cleanest failure-then-mandated-governance arc. The model used a cohort-default-rate feature (a school's aggregate default rate priced into an individual's terms) that disparately impacted Black and Hispanic applicants, and an immigration-status rule that automatically denied certain non-citizen applicants, while the organization ran no disparate-impact testing and gave inadequate adverse-action notices. The remedy did not fine-and-close: it mandated the missing program — model governance, disparate-impact testing, documentation, and reporting controls. The enforcement action wrote the governance the deployment had never built.

Explore this deployment in the PAN Lab →

System map

Who is in the system and what pushes on it

Who is in the system

Agency leadership. Owns procurement, policy, and the authority map; answers for the system publicly.
Served people & families. Those the decisions land on. Deliberately outside the PAN dynamics — their outcomes are measured, never simulated.
Vendors. Build and update the systems; hold the information asymmetry procurement must govern.
Regulators & oversight bodies. Boards, auditors, data-protection officers, inspectorates — external correction capacity.
Courts & commissions. The heaviest, slowest actors — who end most of the failures documented in this Atlas.
Advocates & community organizations. Surface harms institutions do not see; historically the earliest accurate signal.

Dominant pressures

Reviewer bottleneck. One fixed-capacity checking stage sits between AI output and consequence; everything queues behind it.
Austerity & recovery incentives. Cost-cutting and overpayment-recovery targets tilt the system toward denial and enforcement errors.
Vendor opacity. The deploying institution cannot inspect the model, data, or update pipeline it is accountable for.
Data & policy drift. The world, the intake process, and the rules change under a system trained on how things used to be — two mechanisms with different remedies: the statistical properties of what the system processes move (concept drift), or the mixture of inputs arriving in deployment differs from the mixture it was trained on (covariate shift).
Compliance over substance. Paper controls (sign-offs, checklists) satisfy audits while the behavior they describe erodes.

Governance

Questions leaders should be asking

1. Underwriting here is fully automated, so there is no per-application human to catch an error — which means every governable lever is upstream (model choice, testing, the search for alternatives); are those levers actually resourced, or is 'the model decides' treated as the end of the accountability?
2. A denial has to be explained with specific, accurate reasons regardless of how complex the model is — so can the organization and its front line actually tell an applicant why, or has passing the disparity test been mistaken for discharging the separate duty to explain?
3. A facially-neutral aggregate feature — a school's default rate, a ZIP code — can price a group's history into an individual's terms; so is anyone testing the features for protected-class impact, and would a disparity be caught by design or only by a regulator?
4. The hard question is not only whether a disparity exists but how hard the law requires searching for a less-discriminatory model that performs as well — so is that search being done and documented, or is it left to an impasse or an enforcement order to force it?

For the actions behind these questions, see the Practice Library.

Seeing your organization in this domain? Mapping its actual pathways, pressures, and correction capacity is engagement work.

Work With 10²³AI