C-DAG | public financial validation

REPLAY. AUDIT. VERIFY.

Replayable causal audit traces for high-risk financial AI decisions.

Validated across public financial datasets spanning mortgage performance, mortgage applications, consumer complaints, CRT, and CAS disclosure data.

100k+

public financial rows processed

6

validation lanes

5

public financial data types

1.0

replay success across validation runs

governance

HOW IT SOLVES THE GOVERNANCE GAP

01.

Decision trace

Links observed evidence to inferred causal pathway with state-level probabilities.

02.

Counterfactual check

Shows before/after risk deltas under explicit interventions for contestability review.

03.

Replay verification

Deterministically recomputes audit records against active model/policy contracts.

04.

Audit hash-chain

Adds tamper-evident chaining for review workflows and escalation evidence.

05.

Evidence pack

Exports decision outputs, fairness diagnostics, and replay proofs for governance files.

C-DAG proof block

WHAT C-DAG PRODUCES

  • Causal decision pathway
  • Counterfactual risk deltas
  • Deterministic replay result
  • Tamper-evident audit hash
  • Fairness / segment diagnostics
  • Exportable evidence pack

BOUNDARIES

validation proof

PUBLIC FINANCIAL VALIDATION PROOF

Freddie + Fannie + HMDA

rows: 30,000

distribution: APPROVE 9,584 / REVIEW 4,368 / DECLINE 16,048

replay: 1.0

audit-chain: verified

Fannie CAS April 2026

rows: 10,000

distribution: APPROVE 7,948 / REVIEW 686 / DECLINE 1,366

replay: 1.0

audit-chain: verified

Freddie/STACR CRT

rows: 10,000

distribution: APPROVE 9,299 / REVIEW 0 / DECLINE 701

replay: 1.0

audit-chain: verified

CFPB complaints

rows: 10,000

distribution: APPROVE 0 / REVIEW 9,837 / DECLINE 163

replay: 1.0

audit-chain: verified

Baseline outcome holdout

rows: train 58,579 / test 41,421

metrics: AUC 0.573062 / PR-AUC 0.006059

replay: 1.0

audit-chain: verified

loss exposure evidence

LOSS EXPOSURE / BUSINESS RISK EVIDENCE

Public enforcement and risk research show how failures can reach million-to-billion-dollar impact. These cases map directly to C-DAG evidence artifacts used before escalation.

C-DAG LOSS EXPOSURE PACK

CFPB / Wells Fargo order

Public fact: CFPB announced a $3.7B order tied to auto-loan, mortgage, and deposit-account mismanagement.

Workflow risk: servicing control gaps, fee handling, and escalation traceability.

C-DAG artifact fit: causal trace, replay check, audit hash-chain, evidence pack.

CFPB source

FINRA 2025 fine categories

Public fact: AML, communications, trade reporting, and recordkeeping were major recurring fine categories.

Workflow risk: weak surveillance, reporting quality, and records integrity controls.

C-DAG artifact fit: replayable decisions and tamper-evident audit response chain.

FINRA source

SEC AI-washing enforcement focus

Public fact: SEC commentary and enforcement posture continued to target materially misleading AI claims.

Workflow risk: governance claims not backed by traceable decision controls.

C-DAG artifact fit: counterfactual review, replay result, evidence pack for claim support.

SEC source

AI operational loss research

Public fact: published research reports higher operational losses per assets for BHCs with greater AI investment intensity.

Workflow risk: scaling AI activity without matching model/process incident controls.

C-DAG artifact fit: deterministic replay, fairness diagnostics, auditable escalation evidence.

Research source

OCC risk framing

Public fact: OCC risk perspective highlights technology adoption as beneficial but requiring disciplined risk governance.

Workflow risk: model, cybersecurity, and compliance-control gaps in high-risk decisions.

C-DAG artifact fit: causal pathway, policy traceability, audit chain, exportable evidence pack.

OCC source

C-DAG does not prove prevention or savings. It demonstrates how high-risk financial decisions can be made replayable, inspectable, and auditable before issues escalate.