BookbagBookbag
Book Demo
Lending & Credit

Every AI Lending Decision, Audited Against Evidence and Regulation

Ensure AI-driven underwriting, credit scoring, and adverse action decisions are explainable, fair, and regulation-ready.

ECOAFCRACFPB guidanceDodd-Frank §1071OCC Model Risk Management (SR 11-7)Fair Housing Act
Get a Free AuditSee How It Works

The Problem

Lenders are deploying AI for credit decisions at unprecedented scale — automated underwriting, instant credit limit adjustments, dynamic pricing, and adverse action generation. The speed is compelling, but regulators don't care how fast you process applications. They care whether your AI can explain why it denied this applicant while approving that one. When the CFPB examines your lending model and you can't produce decision-level audit trails tied to specific evidence and policy rules, the enforcement action writes itself.

  • AI underwriting models can't produce the adverse action explanations ECOA requires
  • Fair lending analysis is impossible without decision-level evidence trails
  • Model risk management teams can't validate individual AI credit decisions
  • Disparate impact patterns go undetected until regulatory examination
Evidence Payload
evidence
Credit score: 682, DTI ratio: 38%, employment tenure: 3.2 years, requested amount: $285,000, LTV: 87%, payment history: ...
policy_context
Product: 30-year fixed conventional. Minimum credit score: 620. Maximum DTI: 43% (QM safe harbor). Maximum LTV: 95% with...
ai_generated_content
DECISION: APPROVED with conditions — Rate: 6.875% (risk-adjusted +0.375%), PMI required, maximum loan amount: $275,000 (...

What Gets Submitted

What gets submitted when an AI lending decision is audited

evidence
Credit score: 682, DTI ratio: 38%, employment tenure: 3.2 years, requested amount: $285,000, LTV: 87%, payment history: 2 late payments (30-day) in 24 months, assets verified: $42,000 liquid.
policy_context
Product: 30-year fixed conventional. Minimum credit score: 620. Maximum DTI: 43% (QM safe harbor). Maximum LTV: 95% with PMI. Adverse action required per ECOA §1002.9 for any denial. FCRA §615 notice requirements.
ai_generated_content
DECISION: APPROVED with conditions — Rate: 6.875% (risk-adjusted +0.375%), PMI required, maximum loan amount: $275,000 (reduced from requested $285,000 based on asset reserves).
model_trace
Credit pull → score validation → DTI calculation → LTV assessment → risk tier assignment → rate calculation → condition generation → decision output
model_metadata
model: underwriting-v4.1, confidence: 0.89, features_used: 23, adverse_action_codes: [insufficient_reserves], last_validated: 2024-02-28
redacted_fields
SSN, bank_account_numbers, employer_ein, co_borrower_pii

How the Gate Works

Step 1

Submit Evidence

AI decision + evidence payload submitted for structured evaluation

Step 2

Review Against Policy

Decision evaluated against Lending & Credit regulations and policy context

Step 3

Verdict & Audit Trail

Structured verdict with failure categories, corrections, and immutable audit record

Evaluation Taxonomy

Failure Categories

  • Adverse action explanation inadequate
  • Risk pricing not supported by evidence
  • DTI calculation error
  • Fair lending indicator
  • Missing required disclosure
  • Model feature not in approved list

Business Impact

  • CFPB enforcement action
  • Fair lending violation
  • Disparate impact finding
  • Class action exposure
  • Model risk management failure

Evidence Sufficiency

  • Complete underwriting file with all verifications
  • Partial documentation — missing income verification
  • Critical credit data unavailable
  • Evidence contradicts decision rationale

Example Verdict

verdict: needs_fix decision_type: underwriting_decision failure_categories: [adverse_action_inadequate, pricing_unsupported] primary_failure: adverse_action_inadequate severity: high business_impact: cfpb_enforcement_risk EVIDENCE REVIEW credit_score: 682 (above minimum 620) ✓ dti_ratio: 38% (below 43% QM limit) ✓ ltv: 87% (below 95% max) ✓ asset_reserves: $42,000 — covers 3.2 months PITI FINDING "Rate adjustment of +0.375% cites 'risk-adjusted' but does not map to specific adverse factors per ECOA §1002.9. Loan amount reduction from $285K to $275K cites 'asset reserves' but reserves exceed 3-month PITI requirement. Adverse action notice must specify which factors drove conditions." CORRECTED NOTICE "Conditions applied due to: (1) LTV > 80% requiring PMI, (2) credit score in Tier 3 range (660-699) per published rate sheet." AUDIT TRAIL reviewer: sme_lending_3291 reviewed_at: 2024-05-18T11:42:07Z policy_version: underwriting-2024-q2 model_validated: yes (SR 11-7 compliant)

Compliance Frameworks

ECOAFCRACFPB guidanceDodd-Frank §1071OCC Model Risk Management (SR 11-7)Fair Housing Act

Frequently Asked Questions

Related Use Cases

Insurance Claims

Ensure AI-driven claims adjudication, denial rationale, and settlement recommendations are evidence-supported and regulation-compliant.

Learn more

Real Estate

Ensure AI-driven property valuations, tenant screening, and lending recommendations comply with fair housing requirements.

Learn more

Government Benefits

Ensure AI-driven eligibility determinations are fair, documented, and compliant with federal oversight mandates.

Learn more

See how Bookbag audits AI decisions

Join the teams shipping safer AI with real-time evaluation, audit trails, and continuous improvement.

Request a demoGet a free audit