Key insights
- Use AI to speed up review work (summaries, sorting, spotting outliers), not to make compliance decisions.
- Treat AI output like a draft memo: helpful, but it still needs human review and signoff.
- AI is only as good as the records you give it — if your ledgers, support, or tenant files are incomplete, the output won’t be reliable.
- Be ready to explain and document what the tool used, who reviewed it, and how decisions were made — especially for auditors, investors, or HUD reviewers.
Improve AI accountability across HUD and LIHTC programs.
Artificial intelligence (AI) is increasingly embedded in property management software, compliance platforms, and third-party tools, often without formal governance.
For U.S. Department of Housing and Urban Development (HUD) and Low-Income Housing Tax Credit (LIHTC) regulated projects, this creates risk if AI outputs are treated as conclusions rather than inputs.
AI can accelerate analysis and surface trends, but it doesn’t replace professional judgment, internal controls, or accountability.
What HUD is doing with AI
HUD published two foundational documents outlining how the agency intends to use and govern AI:
- HUD AI Strategy
- HUD AI Compliance Plan (issued in response to OMB guidance on federal AI governance)
Together, these documents make clear AI is intended to support, not replace, human judgment.
Accountability remains with HUD personnel, and AI use must be governed through documented risk management and transparency. HUD also highlights appropriate uses such as data review, risk identification, and workflow support, while cautioning against reliance on automated conclusions.
What AI can do in HUD and LIHTC environments
Summarize governing documents
Partnership agreements, regulatory agreements, and similar documents can be summarized to highlight terms that may require follow-up review, such as reserve requirements, guarantees, development fee provisions, and related-party terms.
Identify key debt terms and covenants
Loan documents and debt agreements can be reviewed to highlight reporting requirements, reserve provisions, repayment terms, covenant thresholds, and potential default triggers for management review.
Organize information and supporting document review
Documentation can be organized, key terms extracted, and summaries pulled together to help teams focus their review on higher-risk areas.
Highlight unusual patterns or outliers
Trends, anomalies, or inconsistencies across portfolios can be identified, including unusual expense patterns, occupancy fluctuations, or other items that may warrant closer review.
In each case, AI should support review, while final conclusions remain with the people responsible for it.
What AI can’t do
AI can process large volumes of information efficiently, but it shouldn’t be used to conclude whether:
- Costs are eligible under Section 42 or HUD cost certification requirements
- Costs are allowable under internal policies, regulatory requirements, or contractual agreements
- Tenant data is complete and compliant with program rules
- Financial trends are reasonable in context
- Exceptions are justified or documented
- Layered program requirements are correctly interpreted (e.g., LIHTC combined with HUD rental assistance)
Things to watch out for in LIHTC and HUD contexts
Accuracy and completeness
AI outputs reflect underlying data. If general ledgers aren’t reconciled, cost certification support is incomplete, or tenant records are incomplete, AI won’t be able to provide accurate outputs.
Reasonableness and explainability
Whether evaluating eligible basis, operating deficits, or compliance trends, results must be explainable.
Governance and documentation
HUD places heavy emphasis on governance. Organizations should be prepared to document what data is used, how outputs are reviewed, where judgment is applied, and how errors or bias are identified and addressed.
Questions to ask before using AI outputs
- What data did this output rely on?
- Who reviewed it and what was the conclusion?
- Would I be comfortable defending this to an investor, auditor, or HUD reviewer?
- Where is this documentation retained?
As AI use expands, management should consider adopting a clear internal policy or set of guidelines addressing approved uses, review expectations, documentation, and accountability.
AI may change how information is analyzed, but responsibility for the conclusions remains with people, just as it has with Excel for decades.
How CLA can help with AI risk management
As AI becomes more integrated into HUD and LIHTC programs, CLA supports owners, developers, investors, and property managers in using these tools responsibly.
CLA can help evaluate the accuracy, completeness, and reasonableness of AI-supported information, assess related risks and controls, and align AI use with regulatory, compliance, and financial reporting expectations.
Our focus is on using AI to enhance, not replace, professional judgment, governance, and accountability in affordable housing property management.