NAR's New AI Policy for Brokerages Is Here
The National Association of REALTORS® has released a recommended AI policy for brokerages. Here's what it calls for, and how to get ready before it stops being optional.
NAR just told brokerages how AI should be governed
The National Association of REALTORS® has published a recommended AI policy template for brokerages (opens in new tab), laying out the safeguards it believes should surround everyday AI use: who oversees it, what gets human review, which tools are allowed, and what client data must never touch an AI system.
Today it's a recommendation, and NAR says brokerages should adapt it to their own needs (the template comes in short-form and long-form versions). But the direction is obvious, and NAR is saying it out loud: at the 2026 REALTORS® Legislative Meetings, NAR's chief data officer David Conroy urged brokers to get a policy into agents' hands sooner rather than later, because agents are already experimenting with these tools, and to build a real enforcement plan behind it.
The brokerages that formalize this early set the terms calmly. The ones that wait will do it after an incident, a complaint, or a regulator makes them.
The four risks NAR names
Straight from NAR's own brokerage guidance: these aren't hypotheticals, they show up in everyday workflows.
Accuracy
AI tools hallucinate: wrong square footage, made-up property features, invented market details. If it lands in a listing or client email, the agent and the brokerage are still responsible.
Fair housing
Models trained on biased data produce biased output, like language about who a home is "perfect for" or neighborhood descriptions that drift into steering. Unintentional bias is still a liability.
Privacy
Consumer AI tools can store prompts and train on them. Client financials or transaction documents pasted into an unapproved tool leave the brokerage's control for good.
Licensing
AI can't make agency decisions, give legal advice, or value property on its own. Agentic tools that fire workflows automatically can wander into licensed activity with no one watching.
"Things can get really dicey when your AI tool is acting in similar ways as a licensed activity." David Conroy, Chief Data Officer, National Association of REALTORS®
Eight things the policy calls for
Designate an AI oversight lead
Someone in the brokerage owns AI: which tools are approved, how they're used, and who answers questions before a problem becomes a complaint.
Human review of AI-generated content
Listings, marketing, client communications, transaction documents, and valuations all get human eyes before they go out. The agent stays accountable for anything AI helped produce.
Approved tools only
Personal and consumer-grade AI accounts are flagged as a risk. Client and transaction data belongs in vetted, brokerage-approved tools with audit trails.
Keep sensitive data out
Clear rules about what never gets pasted into an AI system: personal, financial, and transaction details that agents handle every day.
Fair housing compliance
AI-drafted listing descriptions, ads, and chatbot replies get reviewed for discriminatory language, steering, and other violations.
Watch chatbots and agentic AI
Guidance for AI that talks to consumers or acts on its own, including when a conversation must escalate to a licensed human, and where automated workflows edge into licensed activity.
Ongoing AI training
Not a one-time seminar. NAR's guidance calls training as important as the policy itself: agents need to understand the rules and why they exist, as the tools change under their feet.
Enforcement and incident reporting
A policy only counts if it's monitored: defined violations, a clear reporting path, and real follow-through. NAR itself uses a tiered approach, with consequences that scale with the infraction.
Sources: NAR's Brokerage AI Policy Kit, March 2026 (member resource, reviewed in full, summarized here in our own words), NAR's Why Every Brokerage Needs an AI Use Policy (opens in new tab) and AI's Current Iteration Poses Higher Risk and Reward for Brokers (opens in new tab), plus coverage from RE Technology (opens in new tab).
What's actually inside NAR's AI Policy Kit
The March 2026 kit is four documents, and it goes far deeper than most coverage suggests. We've reviewed the whole thing so you don't have to start from a blank page.
The AI Use Policy Template
Twelve sections covering scope, definitions, governance, security and data protection (with dedicated MLS-data and client-data rules), acceptable use, fair housing, chatbots, brand integrity, legal and vendor compliance, supervision and training, incident response, and enforcement.
An AI glossary
Plain-English definitions your agents will actually be held to, from hallucination and performance drift to shadow AI and agentic AI.
An agent & staff FAQ
The questions your agents will ask on day one: which tools are off-limits, listing descriptions, CMAs, virtual staging, social media, and who's responsible when AI gets it wrong. (Answer: the agent, always.)
A one-page summary
The bottom line for the bullpen: humans review everything, approved tools only for client and transaction data, consumer tools only for generic non-sensitive work.
The parts most brokerages can't do alone
Adopting the template is the easy half. These are the obligations it creates:
- Vendor contracts that prohibit AI companies from training on your client data, with due-diligence documentation
- Audit logs of AI usage, plus monitoring for anomalies and bias, not just logging
- MFA and encryption verified on every AI system touching brokerage data
- MLS-data rules: no scraping or repurposing, accuracy review before submission, tool-by-tool approval
- Alignment with recognized frameworks like NIST AI RMF and ISO/IEC 42001
- Training at adoption, at onboarding, and at least annually, with supervision that satisfies state licensing law
- A tiered enforcement ladder from retraining up to termination and licensing-authority referral
Every one of these is a service, a system, or a review we already run. That's the point of the next section.
Support Geeks was built for exactly this
NAR's own implementation guidance says: audit current AI use, assign oversight, approve a tool list, train your agents, and build a review process. Every one of those maps to something we already do, every day.
You need an AI oversight lead
We serve as your fractional AI lead
Most brokerages don't have an AI expert on staff, and don't need one full-time. We fill the role: vetting tools, setting the approved list, answering agent questions, and keeping your policy current as the technology moves.
Explore AI ConsultingYou need a written, enforceable policy
We draft it around how your brokerage actually works
We start from NAR's template and your state's rules, audit how your agents already use AI, and produce a policy your team will actually follow: approved tools, data rules, review requirements, and incident reporting.
Explore AI ConsultingYou need agent training, ongoing
Hands-on workshops built on real estate work
Live sessions built on your agents' real tasks: listings, marketing, follow-up, and what never goes into a chatbot. Everyone leaves with plain-English playbooks, and office hours keep the training current.
Explore AI TrainingYou need compliant tools, not banned ones
We build and vet AI the policy can approve
Prohibition doesn't stick; agents just use personal accounts quietly. We help you stand up approved, auditable tools, and build custom ones where nothing on the market fits.
Explore AI DevelopmentWe don't just talk about AI in real estate. We ship it.
Our incubator built and operates Veroax (opens in new tab), an AI platform that reads real estate disclosure packages and surfaces what matters in minutes. That means we live with the exact problems NAR's policy is about: protecting client data inside an AI system, keeping a human review step where it counts, and documenting how decisions get made.
When we write your AI policy, train your agents, or vet a tool for your approved list, the advice comes from operating production AI in this industry, not from reading headlines about it.
- Veroax - our own real estate AI, in production
- Six products shipped through our incubator
- AI consulting, training & development as core services
- Real estate roots - built by people who work in this industry
NAR AI policy FAQ
Is NAR's AI policy mandatory for brokerages?
No. NAR released it as a recommended, customizable template, not a requirement. But it signals where the industry is heading: brokers formalizing oversight, training, and enforcement before regulators or lawsuits force the issue. Adopting a policy early is far cheaper than reacting later.
What does NAR's AI policy template actually cover?
NAR's March 2026 Brokerage AI Policy Kit includes a 12-section policy template (long-form plus a one-page version), an AI glossary, and an agent FAQ. The policy covers governance and an AI Oversight Lead, mandatory human review of AI content before it reaches a listing, client, marketing, transaction document, or CMA, approved-tools rules that ban consumer-grade accounts for client data, dedicated MLS-data and client-data protections, fair housing requirements with concrete prohibited-language examples, chatbot and agentic-AI limits, vendor contract requirements, annual training, incident response, and tiered enforcement.
What is agentic AI, and why is NAR worried about it?
Agentic AI doesn't just draft content, it acts: sending messages, updating CRM records, scheduling follow-ups, and triggering workflows without a human in the loop. NAR's chief data officer has flagged the risks for brokerages: acting on stale or non-compliant data, automated messaging that drifts into fair housing exposure, off-brand or hallucinated marketing, and workflows that cross into activity requiring a real estate license. If your brokerage uses AI that acts on its own, your policy needs to say where a licensed human must take over.
Where do I get the actual NAR template?
It's available to NAR members at nar.realtor under the AI Policy Template for Brokers. We work from it, plus your state's requirements, when we draft a brokerage's policy.
Why is Support Geeks qualified to help with AI compliance?
We don't just advise on AI, we build and operate it in real estate. Our incubator created Veroax, an AI platform that analyzes real estate disclosure packages, so we deal with client data handling, accuracy review, and fair-housing-adjacent risk in our own product every day. That experience goes into every policy, training, and tool we deliver.
Is this legal advice?
No. We help with strategy, policy drafting, training, and tooling; your attorney should review anything with legal effect. We're happy to work alongside your counsel.
Support Geeks provides strategy, training, and technology services, not legal advice. Have your attorney review any policy before adoption.
Get ahead of the AI policy wave
Book a free readiness consult. We'll walk through NAR's recommendations against how your brokerage uses AI today, and show you exactly what we'd put in place.