The risks and rewards of AI in commercial real estate
AI can make your CRE operation faster and leaner. It can also make you liable. The difference is your process: where you use it, how you supervise it, and what you document.
Key takeaways
- Clear wins: analytics, smart maintenance, lease drafting support, market research, expense optimization, and chatbot triage.
- High-risk zones: tenant/employee screening, credit checks, and lease negotiations without human control.
- Compliance requires bias testing, audit trails, and a human in the loop.
- Legal risks include copyright exposure from images and AI-fueled voice scams tied to wire transfers.
Where AI helps right now
Use predictive analytics to forecast vacancies, rent trends, and OpEx. Plug smart maintenance into HVAC, elevators, and building systems to spot issues before failure.
Deploy chatbots for basic prospect questions, lead routing, and appointment setting. For lease prep, let AI compare forms and flag missing exhibits or cross-references so your team spends time on substance, not clerical work.
What you cannot outsource to AI
"If you couldn't do before you began using your AI product, you can't do it now," said Katarina "Kate" Polozie, partner at Woods Oviatt Gilman. This is critical for tenant, credit, and employee screening. "You can be liable for what your chatbot provides."
AI outputs can sound convincing and still be wrong. As Polozie put it, "They sound real, you want to believe it, but they may not be real." Keep a human in the loop for any decision that affects housing, employment, or credit.
Screening and compliance checklist
- Run regular bias tests on your screening models; document methodology, results, and remediation.
- Use clear adverse action workflows and required notices under applicable laws (e.g., fair housing, employment, and credit).
- Log data sources, model version, prompts, and decisions for auditability.
- Train staff: AI suggestions are not decisions. Final calls must be made by qualified humans.
- Review relevant guidance from regulators such as the EEOC on AI use in hiring and screening (EEOC AI resources).
Leasing workflows: use AI for prep, not for leverage
AI is useful for lease drafting support. It can compare competing forms, spotlight pitfalls, and catch missing exhibits, said Kris Vurraro, partner at Woods Oviatt Gilman with a concentration in commercial leasing.
But it's a poor negotiator. "AI lacks any contextual knowledge," Vurraro said. It doesn't know your client's goals, risk tolerance, or leverage. Use it to prepare your playbook; keep humans on the calls and at the table.
Images and copyright: don't guess, don't scrape
Listings live or die on photos. Don't rely on AI-generated or unvetted images. You may not know the source or ownership, which can expose you to claims.
Use licensed photography, in-house shoots, or permissions you can prove. Archive release forms and licenses with each listing.
Wire fraud and AI voice cloning
Wire transfers are a prime target. Polozie warned that a 30-second clip is enough for a realistic voice clone. If you get a call changing wire instructions, assume it's suspect.
- Use out-of-band verification (call a known, verified number; never the number in the message).
- Require dual approval for any change to payment or account details.
- Set a no-exceptions policy: no wire changes by voice alone.
- Educate clients and counterparties before closing week.
- See guidance on voice cloning and imposter scams from the FTC (FTC advisory).
Vendor due diligence: privacy, models, and terms
Know how your tool works today-and how it could change. Terms of use can shift. Data "not retained" now may be retained later.
Vurraro's reminder: "Paid does not mean private in the way you want it to." Validate whether the model is closed or public, how your data is used, and if your prompts/outputs can train future models.
Procurement questions to ask
- Data handling: retention, deletion, training usage, and who can access your data.
- Model type: closed vs. public; ability to run in a private tenant.
- Security: SOC 2/ISO attestations, incident response, and audit logs.
- Compliance: fair housing/employment testing features and reporting.
- Contract: notice periods for ToS changes and opt-outs for data usage.
- Controls: role-based access, redaction, and watermarking for outputs.
- Legal review before rollout; document your governance plan.
Quick-start playbook for brokers and owners
- Pick 1-2 low-risk pilots: maintenance alerts, market comp summaries, or lease exhibit checks.
- Define "AI can" vs. "AI cannot" rules for your team. Decisions stay human.
- Bias-test any screening workflow before and after launch; schedule quarterly retests.
- Adopt a wire protocol with out-of-band verification and require client acknowledgment.
- Centralize prompts, templates, and approvals to standardize how AI is used.
- Track ROI (hours saved, errors caught) and risk metrics (exceptions, incidents).
Bottom line
Use AI to compress busywork and surface risks. Keep people in charge of choices that carry legal or financial consequences. Small, controlled wins compound-reckless automation does not.
If your team needs structured upskilling with governance in mind, explore role-based programs at Complete AI Training.
Your membership also unlocks:
