FCRA · ECOA · CFPB Circulars 2022-03 / 2023-03 / 2024-04

Specific reasons.
Cryptographically signed.
Per decision.

CFPB Circulars 2022-03 / 2023-03 / 2024-04 + Reg B § 1002.9(b)(2) compliance, without the 0.5–1 FTE. AttestProto signs each adverse-action decision and auto-maps the citation, 30-day notice, and human-review pathway.

Generate one signed decision →View on GitHub
the cost of the gap

Compliance VPs spend 0.5-1 FTE rebuilding adverse-action paperwork that should be automatic.

Specific-reasons disclosure

Reg B § 1002.9(b)(2) requires the principal reasons for adverse action. "AI denied" doesn't qualify. CFPB has said so three circulars in a row.

30-day notice + human review

Section 1002.9(a) requires written notice within 30 days. Mapping the AI decision to a human-reviewable record is hand-rolled at most fintechs.

Cross-jurisdiction overlay

Add Colorado AI Act, GDPR Article 22 (EU customers), state UDAP analogues. Your team rebuilds the same compliance bundle four ways.

how it works

Sign the decision. Auto-map to the citation. Done at decision time, not month-end.

# underwriter side — every credit decision $ attestproto sign \ --agent credit-model-v7 \ --tool decision.adverse \ --input '{"applicant_id":"...","factors":[...],"score":612}' \ --out adverse-2026-05-05-a-2731.json # compliance side — auto-map to citations $ attestproto map \ --rules fcra-1681m,ecoa-reg-b-1002.9,gdpr-art-22,cfpb-circ-2022-03 \ --bundle adverse-q1.tar.gz # 1,294 decisions · 1,287 mapped clean · 7 missing specific-reasons → flagged
CFPB Circular 2022-03Creditors cannot justify noncompliance with ECOA based on the mere fact that the technology they use to evaluate credit applications is too complicated or too opaque to understand.
Reg B § 1002.9(b)(2)The notification given to an applicant when adverse action is taken shall be in writing and shall contain a statement of specific reasons for the action taken.
try it

Sign one adverse-action decision in your browser.

No install, no account. The keypair is generated client-side and never leaves this page. Same code path as the production CLI.

Generates an Ed25519 keypair in your browser. Nothing leaves this page.
v0.1.1 — auto-generated notices

Adverse-action notice rendered from the attestation. Compliant out of the box.

Every signed adverse-action attestation can render a customer-facing FCRA § 1681m + ECOA Reg B § 1002.9(b)(2) notice — specific reasons, ECOA rights statement, FCRA consumer-reporting-agency disclosure, all in the language CFPB examiners expect. Plain text or Markdown. One CLI flag.

# after sign_attestation produced adverse-2026-05-06.json $ attestproto adverse-action adverse-2026-05-06.json \ --creditor-name "Upstart Holdings, Inc." \ --creditor-address "2950 S Delaware St, San Mateo, CA" \ --creditor-phone "1-855-438-8778" \ --applicant-name "Jane Q. Applicant" \ --application-id "APP-2026-00123" \ > notice.txt # → ECOA + FCRA + CFPB Circulars 2022-03 / 2023-03 / 2024-04 satisfied

Pulls «specific reasons» from attestation.output.factors directly — no per-notice copywriting. Markdown output drops into your audit bundle unchanged.

insurance + liability

Does your current policy actually cover an AI lending decision claim?

Most AI lending operators discover too late that their cyber, generic E&O, or D&O policy is silent or excludes algorithmic discrimination. Pick your current policy stack — see how it actually responds to the eight liability scenarios CFPB, state AGs, and class-action plaintiffs are filing in 2024-2026. Underwriters increasingly require per-decision attestation evidence on renewal.

Pick your current policy stack

E&O with explicit AI / algorithmic decision-making endorsement (Munich Re, Vouch, Coalition specialty)

Liability scenarioCoverage under E&O — AI/algorithmic endorsementStatute / case basis
FCRA class action — bad reason codesCovered15 U.S.C. § 1681m + § 1681n / § 1681o
ECOA class action — adverse action notice failureCovered15 U.S.C. § 1691 / Reg B § 1002.9
CFPB enforcement — explainability deficitPartialCircular 2022-03 + UDAAP
State AG action (NY DFS, CO Atty Gen, CA AG)CoveredNY 23 NYCRR 500 / CO AI Act / CCPA
EU AI Act high-risk fine (incoming)PartialRegulation (EU) 2024/1689 Art. 99
PII breach in attestation logsSilent (assume excluded)GLBA Safeguards / state breach laws
Sub-vendor (LLM provider) liability spilloverPartialCFPB 2024-04 + service-provider doctrine
Plaintiff alleges no adverse action notice ever sentCoveredFCRA § 1681m(a) + private right of action
Underwriter perspective: The right answer for AI lending. Endorsement adds ~15-25% premium. Underwriters require attestation evidence (per-decision logs, bias audits) — AttestProto satisfies the typical evidence schedule.
How to read this:‘Silent’ means the policy form doesn't address the scenario explicitly — courts will typically rule against coverage if the underlying conduct involves an AI-generated decision and the form pre-dates 2023. Force the carrier to give you a written coverage opinion in advance — not on day-of-claim. AttestProto's per-decision evidence is what most carriers ask for during underwriting and discovery.

Not legal or insurance advice. Coverage varies by carrier, jurisdiction, endorsement, and policy form. Validate with your broker before relying.

Take this matrix to your broker before renewal. If the broker can't answer in writing whether the policy responds to FCRA / ECOA / CFPB claims arising from algorithmic decisions, you don't have coverage — you have hope.

vendor due-diligence

Auditing an AI vendor? Generate the 47-question checklist in 30 seconds.

Pick your deployment context. Get back a context-tailored DD questionnaire mapped to FCRA, ECOA, NYC LL144, CO AI Act, EU AI Act, GDPR Art. 22, and CFPB Circulars. Each question includes the why, the citation, and the evidence to demand. Copy as markdown or download — bring it to your next vendor pitch and watch how many can't answer.

Pick your AI deployment context
17 questions
0/17 marked complete · 0% done

Model + training

Decision attestation

Adverse-action notices

Vendor governance

Data + privacy

Explainability + appeal

EU AI Act

Compiled from FCRA, ECOA, NYC LL144, CO AI Act, EU AI Act, GDPR Art. 22, CFPB Circulars 2022-03 / 2023-03 / 2024-04, OCC Bulletin 2021-39, NY DFS 23 NYCRR 500. Not legal advice — validate scope with counsel for your jurisdiction.

deployment

Self-hosted. Your loan-application data never leaves your infrastructure.

On-prem or VPC

Run inside your existing compliance perimeter. No vendor SaaS round-trip with PII.

Append-only ledger

Hash-chain integrity. Decision history is tamper-evident; regulator examination is straightforward.

4-8× ROI vs hand-rolled

0.5-1 FTE saved at fully-loaded $150-200k vs $25-50k/year licence.

Worth a 20-minute screen-share?

We'll demo the auto-mapping fire on a sample lending attestation. If it doesn't fit your stack, no follow-up.

Contact us →Other use cases