Fair housing compliance for affordable & supportive housing
Every screening decision, lawful and defensible.
FairAudit is AI-assisted compliance infrastructure for property managers of affordable, supportive, and homeless housing. Every applicant receives due process — and every determination stands up to a fair-housing investigator, an auditor, or a court.
- 13
- integrated compliance modules
- 8
- due-process safeguards mapped to law
- 3
- challenge pathways: accuracy, relevance, mitigation
Application case file
D. Alvarez
Pending reviewHarbor Light Supportive Housing · HUD-VASH voucher
- Consumer report data ingested
- Stale eviction filing quarantined — no disposition
- Individualized assessment completed
- Adverse-action notice delivered
Impact ratio 0.74 — burden-shifting analysis opened
9f2c…a41e3b7d…0c9fIllustrative product view with fictional sample data
Operationalizes the law that governs your screening
Fair Housing Act
42 U.S.C. § 3601 et seq.
Fair Credit Reporting Act
15 U.S.C. § 1681 et seq.
HUD criminal-records guidance
HUD OGC, Apr. 4, 2016
Section 504
Rehabilitation Act of 1973
Violence Against Women Act
34 U.S.C. § 12491
Source-of-income laws
e.g., Cal. Gov. Code § 12955
Why it matters
Affordable housing runs on screening. Screening runs on legal exposure.
For households leaving homelessness, rebuilding credit, or holding a voucher, a single unexplained denial can mean another night without housing. For operators, the same denial can become a fair-housing complaint with no record to answer it.
Automated scores, human consequences
Screening vendors return opaque scores and raw records. When a facially neutral criterion screens out voucher holders or people with old, dismissed charges, the housing provider — not the vendor — answers for the discriminatory effect.
Program rules layered on civil-rights law
LIHTC, Housing Choice Vouchers, HOME, and Continuum of Care funding each add obligations — Housing First commitments, VAWA protections, Section 504 — on top of the Fair Housing Act and state law.
Records reconstructed after the complaint
When a fair-housing agency, auditor, or court asks why an applicant was denied, most operators rebuild the story from email. By then the reasoning, the notice, and the chance to cure are gone.
Algorithmic tenant screening is now a live civil-rights question: in Louis v. SafeRent Solutions (D. Mass.), voucher holders challenged an automated screening score under the Fair Housing Act — a case that settled in 2024.
The platform
One system of record for the entire screening lifecycle.
Thirteen integrated modules take an application from raw consumer report to a reasoned, reviewable, and preserved determination — and keep watch over outcomes across the whole portfolio.
Adjudicate
Decide on a written standard
Versioned screening policies
Published criteria per property, with guardrails that flag unlawful or high-risk rules as they're drafted.
Data intake & accuracy
Normalize every vendor, match identity, and quarantine stale, mismatched, or disposition-less records.
Relevance to tenancy
Each record is tested for a genuine nexus to tenancy before it may count against anyone.
Human review queue
Borderline cases route to a qualified reviewer; overrides require a written justification.
Protect
Guarantee the applicant's rights
Individualized assessment
The four-factor review HUD guidance calls for before any criminal-history-based denial.
Challenge docket
Applicants dispute accuracy, relevance, or offer mitigation — and receive a written resolution.
Reasonable accommodations
Requests tracked through the interactive process without ever recording a diagnosis.
Adverse-action notices
FCRA-compliant notices with principal reasons, agency disclosures, and dispute rights.
Monitor
Measure outcomes, not intentions
Disparate-impact testing
Approval rates by protected class against the four-fifths benchmark, with sample-size floors.
Burden-shifting analysis
A structured record of necessity and less discriminatory alternatives for every flagged disparity.
Proxy & drift detection
Catch features that stand in for protected traits, and practice drifting from written policy.
AI governance
Scorecards, model cards, and algorithmic impact assessments aligned to the NIST AI RMF.
Preserve
Keep the administrative record
Append-only audit trail
Every determination, review, override, and notice — attributed and time-stamped.
Evidence vault
Notices and assessments sealed with SHA-256 hashes to establish chain of custody.
Jurisdiction overlays
Federal, state, and local rules layered per property; the most protective standard controls.
Tenant isolation
Every query scoped to your organization; audit and evidence tables reject edits and deletions.
How it works
From application to a determination you can defend.
FairAudit sits between your screening vendors and your final decision. It doesn't replace your staff's judgment — it structures it, documents it, and makes it consistent.
- 01
Publish a written standard
Adopt versioned, property-specific criteria. FairAudit flags source-of-income screens, blanket criminal bans, and other high-risk rules before they take effect.
- 02
Test every record
Vendor data is normalized, matched to the right person, and labeled for relevance. Arrests without convictions, sealed records, and stale filings never reach the decision.
- 03
Adjudicate with reasons
The engine applies the policy and states a reason code for every adverse factor. Anything requiring judgment goes to a person — never to a black box.
- 04
Give notice, hear the challenge
Applicants receive the principal reasons and their rights, and can contest accuracy, relevance, or offer mitigation. Each challenge ends in a written resolution.
- 05
Monitor and preserve
Outcomes are tested for disparate impact across protected classes, drift is flagged daily, and the complete record is preserved for audit or litigation.
Legal framework
Procedural fairness, engineered into every decision.
Each safeguard traces to a specific source of law. The result is equal access to housing in practice — not just in policy — and an evidentiary record that shows it.
- Notice of reasonsPlain-language reason codes, each tied to a published criterionFCRA § 615(a), 15 U.S.C. § 1681m(a)
- Maximum possible accuracyIdentity matching and quarantine of mismatched or stale recordsFCRA § 607(b), 15 U.S.C. § 1681e(b)
- Opportunity to be heardAccuracy, relevance, and mitigation challenges resolved in writingHUD OGC Guidance on Criminal Records (2016)
- Individualized assessmentFour-factor review before any criminal-history-based denialHUD OGC (2016); Cal. Code Regs. tit. 2, § 12264 et seq.
- Reasonable accommodationInteractive process tracked from request to resolution42 U.S.C. § 3604(f)(3)(B); Section 504
- Survivor protectionsMitigation pathway for records arising from domestic violenceVAWA, 34 U.S.C. § 12491
- Source-of-income neutralityGuardrails against screening out voucher holdersCal. Gov. Code § 12955; state & local SOI laws
- Discriminatory-effects reviewFour-fifths screening plus a three-step burden-shifting recordInclusive Communities, 576 U.S. 519 (2015)
Citations identify the legal sources each feature is designed around. Applicability depends on the property, program, and jurisdiction; consult counsel.
Who we serve
Built for the housing where the stakes are highest.
Low-barrier and publicly assisted housing serves the people most exposed to screening errors — and carries the densest web of program rules and civil-rights obligations.
Affordable housing operators
LIHTC, HOME, and project-based Section 8 portfolios that need uniform, well-documented screening across dozens of properties and site teams.
Supportive & homeless housing
Continuum of Care and Housing First programs that must keep barriers low while documenting every denial and accommodation.
Housing authorities & voucher programs
PHAs and partners serving Housing Choice Voucher, HUD-VASH, and Emergency Housing Voucher households.
Asset managers & counsel
Compliance leaders, investors, and outside counsel who need portfolio-wide visibility and a record that holds up.
AI that assists. People who decide.
Algorithmic accountability is the product, not a disclaimer. Determinations come from a deterministic engine applying your written policy; AI only drafts, and a person always approves.
- Demographic data (race, sex, familial status, disability) monitors outcomes — it never scores applicants
- Every AI suggestion is reviewable and never self-executing
- Model card and algorithmic impact assessments aligned to the NIST AI RMF
- Proxy detection for features that stand in for protected characteristics
A record built to be relied on.
Compliance evidence is only as good as its integrity. FairAudit is designed so the record can't be quietly rewritten after the fact.
- Organization-scoped data isolation enforced on every query
- Audit and evidence tables reject updates and deletions at the data layer
- SHA-256 content hashes on every preserved notice and assessment
- Encrypted in transit (TLS/HSTS) with attributed, time-stamped actions
Questions
Frequently asked questions
Does FairAudit make tenant-screening decisions?+
No. FairAudit applies the written policy you publish, explains the result, and routes anything requiring judgment to a qualified person on your team. It is a compliance and accountability layer — your staff remain the decision-makers.
Is FairAudit a consumer reporting agency?+
No. FairAudit does not furnish consumer reports. It organizes and evaluates the reports your screening vendors provide, and generates the adverse-action notices the Fair Credit Reporting Act requires you to send.
How is AI used?+
Assistively and never autonomously. AI can draft structured criteria from a plain-language policy or suggest features that may act as proxies for protected traits. Every AI output requires human review before it has any effect, and demographic data such as race or disability is never an input to a determination.
Which laws and jurisdictions are covered?+
The federal Fair Housing Act, FCRA, HUD guidance, Section 504, and VAWA, with a California overlay (FEHA, source-of-income protection, and criminal-history regulations) configured by default. Additional state and local rule sets can be layered per property.
Does FairAudit replace legal counsel?+
No. FairAudit operationalizes legal standards and preserves the record, but it does not provide legal advice. We recommend reviewing policies and any narrowing of the governing standard with counsel.
Can we try it without real applicant data?+
Yes. Every new workspace can load a sample portfolio — a LIHTC family property, permanent supportive housing, and a senior community — with applicants, decisions, challenges, notices, and civil-rights analytics.
Housing is a right worth getting right.
Create a workspace in minutes and explore FairAudit with a sample affordable-housing portfolio — no applicant data required.