Real estate and PropTech compliance

Real estate compliance — Fair Housing Act, ECOA, FCRA, RESPA, TILA, HMDA, state broker rules, AVM rules, state landlord-tenant law.

Real estate AI compliance centers on anti-discrimination + consumer credit law + state real-estate licensing + landlord-tenant law. The protected-class surface is wide and active enforcement is the norm.

Regulatory frame

Regime	Applies when	Evaluation-evidence shape
Fair Housing Act (42 USC § 3601-3619)	Housing transactions, advertising, financing	Protected-language hard rule; disparate-impact testing
ECOA (Reg B)	Consumer credit decisions	Adverse-action notice; bias audit
FCRA (15 USC § 1681)	Consumer reports	Accuracy, dispute rights, adverse-action notice
RESPA / Reg X	Mortgage settlement	Disclosure timing; affiliate-relationship transparency
TILA / Reg Z	Mortgage disclosures	APR, costs, loan-estimate accuracy
HMDA / Reg C	Mortgage originators above threshold	Demographic data collection and reporting
PAVE task force / proposed AVM rule	AVMs in lender mortgage decisions	Bias testing, governance, randomized sampling
CFPB authority	Consumer-finance overlap	UDAAP scrutiny on AI
State broker licensing	Licensed salespeople / brokers	Supervisory evidence
State landlord-tenant law	Residential rentals	Habitability, just-cause, security deposit
Source-of-income, criminal-record, ban-the-box laws	State / local	Per-jurisdiction screening rules
Privacy laws (CCPA, GDPR, etc.)	Per-jurisdiction	DSR, consent, retention
NAR rules / MLS rules	NAR-affiliated brokers	Code of Ethics; MLS-specific rules

Fair Housing Act

42 USC § 3601-3619. Prohibits discrimination on the basis of race, color, religion, sex (including sexual orientation and gender identity per HUD), disability, familial status, national origin in:

• Sale, rental, financing of dwellings
• Advertising of dwellings
• Steering, blockbusting, redlining

HUD has specifically focused on AI-driven listing language and tenant screening.

Stratix evaluation evidence:

• Protected-language hard rule + judge for indirect references
• Steering-language judge for neighborhood characterization
• Disparate-impact monitoring on tenant screening, AVM, recommendation surfaces

ECOA / Regulation B

For credit decisions in mortgage and consumer-finance:

• No discrimination on FHA bases plus age, marital status, public-assistance income
• Adverse-action notices required with specific reasons
• Disparate-treatment and disparate-impact theories both apply

Stratix evaluation evidence:

• Reason-code rule on every adverse output
• Group-fairness scorer
• Audit trail of inputs / model version / decision

FCRA

For tenant screening companies and others producing consumer reports:

• Accuracy obligations (maximum-possible-accuracy standard)
• Consumer dispute rights with reinvestigation
• Adverse-action notice on use to deny housing or credit
• Permissible-purpose limits

FTC and CFPB have enforcement authority and have used it.

Stratix evaluation evidence:

• Accuracy rule per item
• Adverse-action triggering rule
• Dispute-routing rule for tenant disputes

PAVE task force / AVM proposed rule

PAVE (Property Appraisal and Valuation Equity) task force findings drive an active rule-making track:

• Federal banking regulators + CFPB + FHFA proposed AVM rule under FIRREA Title XI
• Quality control standards include nondiscrimination
• Random-sample testing
• Conflict-of-interest controls

Stratix evaluation evidence:

• Demographic-fairness scorer on AVM outputs
• Random-sample audit pipeline
• Governance documentation per AVM use

RESPA / TILA / HMDA

For mortgage origination:

• RESPA — settlement-cost disclosure, anti-kickback
• TILA — APR, finance-charge accuracy
• HMDA — demographic data collection, public reporting

Stratix evaluation evidence:

• Disclosure-content rule per applicable form
• HMDA data routing rule

CFPB UDAAP

CFPB has used UDAAP (Unfair, Deceptive, Abusive Acts or Practices) authority to investigate AI in consumer finance. Particularly:

• Algorithmic discrimination as UDAAP
• "Black-box" decision-making lacking adverse-action specificity
• Dark patterns in consumer-finance funnels

State broker licensing

State real estate commissions regulate licensees. AI tools used by licensees do not exempt licensees from supervisory and disclosure obligations. State-by-state:

• Required disclosures the AI cannot draft autonomously
• Continuing-education requirements that touch AI use
• Discipline for misrepresentations whether AI-generated or not

State landlord-tenant law

Massive variance. Examples:

• CA — strict just-cause eviction; rent control in many cities; security-deposit returns within 21 days
• NY — strong tenant-rights regimes in NYC; rent-stabilized stock
• WA — just-cause statewide; specific source-of-income protections
• TX — landlord-friendly default; specific habitability rules

Stratix evaluation evidence:

• Jurisdiction-aware tenant-rights rule
• Habitability auto-escalate rule

Source-of-income, criminal-record, ban-the-box laws

State and local laws affecting tenant screening:

• Source-of-income protections in 100+ jurisdictions (cover Section 8, SSDI, etc.)
• Criminal-record limits (HUD 2016 guidance binding for federally-funded housing; state/local stronger)
• Ban-the-box analogs for housing

Privacy laws

CCPA / state privacy / GDPR — same framework as retail compliance.

NAR rules / MLS rules

For NAR-affiliated brokers:

• NAR Code of Ethics
• Multiple-Listing Service rules (varies by MLS)
• Recent class-action settlement (commission rules) creates new obligations

Recommended setup

1. Pro tier minimum; Enterprise for brokerages, screening companies, lenders
2. SSO + RBAC for licensee / supervisor / compliance roles
3. Hard rules: protected-class language, agent/loan-officer review gate, FCRA accuracy, jurisdiction-aware tenant rights
4. Group-fairness scorers on every consumer-affecting AI output
5. PAVE-aligned random sampling for AVMs
6. Audit retention matching the longer of state broker / FCRA / fair-housing requirements (default 5 years)
7. State-form refresh at least quarterly
8. DPA executed; tenant data residency where contracted

See also

• Real estate scenarios
• Real estate eval patterns
• Concept: Multi-tenancy
• Enterprise