Key takeaways
- Nearly half of REALTORS use AI daily (23%) or weekly (25%), and only 12% have no plans to use it[1].
- The biggest risks are old rules applied to new tools: fair housing, privacy, truthful advertising, fake reviews and calling consent.
- Clients want to know: 76% of Americans say it is extremely or very important to be able to tell whether content was made by AI or people[2].
- Gallup found 44% of employees say their organization has begun integrating AI but only 22% say it has a clear plan[3]; a written AI policy closes that gap.
AI is already in the office
AI is no longer a pilot project in real estate. In NAR's 2026 technology report, 23% of REALTORS use AI daily and 25% weekly, 31% experiment with it occasionally, 9% are curious but have not tried it, and only 12% say they are not using it and have no plans to[1]. The top use is writing listing descriptions, cited by 75% of AI users[1].
- Daily 23%
- Weekly 25%
- Occasionally 31%
- Curious, not tried 9%
- Not using, no plans 12%
Source: National Association of REALTORS®, 2026[1]
Show the data
| Item | Value |
|---|---|
| Daily | 23% |
| Weekly | 25% |
| Occasionally | 31% |
| Curious, not tried | 9% |
| Not using, no plans | 12% |
Clients notice too. Among sellers who hired an agent, 21% knew and 36% suspected their agent was using AI tools[4]. The compliance question is not whether to use AI, but whether the brokerage knows which tools agents use, for what, and with which client data.
The risk map
| AI use | Main compliance risk | Control |
|---|---|---|
| Listing descriptions and ads | Inaccurate claims; descriptions of who should live there | Fact check against the file; remove people descriptions |
| Virtual staging and image editing | Misleading images | Label staged photos; keep an unedited set |
| Ad targeting and audience tools | Excluding people by protected characteristics | Use housing ad categories; no demographic targeting |
| Chat assistants answering buyers | Steering or unequal answers about neighborhoods | Approved answers; route neighborhood opinion questions to data sources |
| Drafting emails and texts | Deceptive subject lines; missing opt-outs | Send through approved platforms with footers |
| Outbound AI voice calls | TCPA artificial voice consent | Prior express written consent before telemarketing calls |
| Summarizing documents and calls | Client data sent to vendors; errors in summaries | Approved vendors only; human review of anything relied on |
| Review responses and generation | Fake or AI-written reviews | Never generate reviews; disclose any connections |
Fair housing comes first
The Fair Housing Act prohibits discrimination because of race, color, national origin, religion, sex, familial status and disability, and it covers renting, buying, mortgages and other housing-related activities[5]. AI tools can break it without anyone intending to. A text generator trained on old listings may describe a home as ideal for a young couple. A chat assistant may answer a question about safe areas with an opinion that tracks demographics. An ad tool may optimize delivery toward some groups and away from others.
- Ban prompts and outputs that describe the ideal buyer or tenant; describe the property instead.
- Give chat tools approved answers for neighborhood, school and safety questions that point to public data sources.
- Review samples of chat transcripts each month for unequal answers or steering language.
- Use the housing special ad category on platforms that offer one, and avoid lookalike audiences built from client lists.
- Treat screening tools for rentals with extra care, and have counsel review any automated criteria.
Privacy and confidential information
Pasting a seller's disclosure, a buyer's pre-approval letter or a negotiation email into a public AI tool can send confidential client information to a vendor with unclear data terms. Agents rarely think of it as sharing. Set a simple rule: client documents and personal information go only into tools the brokerage has approved, under contracts that limit use of the data, and never into personal accounts. Add each approved tool to the brokerage's data map so privacy requests and breach responses cover it. Our guide to consumer data privacy laws covers the rest of that program.
Truthful content and honest reviews
AI copy is held to the same truth standard as any other ad. A generated description that adds a feature, smooths over a defect or invents a nearby amenity is the agent's misrepresentation. Images need the same care: virtual staging and object removal must not mislead buyers about the property. See Article 12 on truth in advertising for a checklist.
Reviews are a sharper line. The FTC's final rule bans businesses from creating or selling fake reviews and testimonials, including AI-generated fake reviews[6]. Using AI to draft a review for a client to post, or to generate testimonials for a website, is exactly what the rule targets.
Calls, voices and consent
In February 2024 the FCC ruled that calls made with AI-generated voices are “artificial” under the TCPA[7], so AI voice telemarketing calls need the same prior express written consent as prerecorded calls. Answering calls that people choose to make to the brokerage is a different situation from dialing out, and many offices use an SI Employee such as Ava for inbound calls and showing requests. Even then, tell callers they are speaking with an automated assistant, follow your state's recording consent rules, and give them an easy way to reach a person. Our article on AI voice calls and the TCPA covers outbound rules in detail.
Wrong answers are a liability
Generative tools produce fluent text whether or not it is correct. In real estate that shows up as an invented HOA fee, a wrong school assignment, a misread inspection summary or a confident answer about zoning that nobody checked. When an agent forwards that output to a client, the error becomes the agent's statement. Set a firm rule for anything a client might rely on: numbers, dates, legal terms, property features and summaries of documents are checked against the source before they leave the office. For chat and voice tools, configure them to say they will have an agent confirm details rather than guess, and test them with the awkward questions buyers actually ask.
Keep records of AI-assisted work
If a dispute arises, the brokerage may need to show what a client was told and how. Save approved AI-drafted ads with the listing file, keep chat and call transcripts for the same period as other client communications, and record which tool produced a document summary that an agent relied on. Those records also make the monthly review possible, and they show a regulator that AI use was supervised rather than left to individual agents.
Tell people when AI is involved
Disclosure is becoming an expectation even where no law requires it. 76% of Americans say it is extremely or very important to be able to tell whether content was made by AI or people[2], and 50% are more concerned than excited about the increased use of AI in daily life, up from 37% in 2021[2]. A short, plain disclosure costs little and protects trust: a line on chat widgets saying the assistant is automated, a note on virtually staged photos, and an honest answer whenever a client asks.
Example disclosure: “You're chatting with our automated assistant. It can answer listing questions and set up a showing request. An agent will follow up, and you can ask for a person at any time.”
Governance in six steps
Inventory
Ask every agent and staff member which AI tools they use, for what, and with what data. Expect surprises.
Approve
Choose approved tools for each use, based on data terms, security and accuracy, and say which uses are not allowed.
Write the rules
Put a one-page AI policy in the manual: approved tools, banned data, review requirements, disclosure wording and who to ask.
Review output
Require human review before AI content is published or sent, and sample chat and call transcripts monthly.
Train
Run a short session with real examples from your office: a description that needed fixing, a chat answer that drifted.
Revisit
Review the policy every six months; tools and rules change quickly.
This is where most organizations lag. Gallup found that 44% of US employees say their organization has begun integrating AI, but only 22% say it has communicated a clear plan[3]. A brokerage with a written policy and a review routine is ahead of most.
Questions to ask AI vendors
- Will our client data be used to train your models or shared with others, and can we turn that off?
- Where is data stored, for how long, and how do we delete it?
- Can we see and export transcripts or logs for review and record keeping?
- How does the tool handle neighborhood, school and safety questions?
- Does any feature place outbound calls or texts, and how is consent checked?
- What happens when the tool does not know an answer?
Example: an AI policy that caught a problem
Example (fictional): A brokerage adopts a one-page AI policy and asks agents to send their first five AI-drafted listing descriptions to the managing broker. One draft describes a condo as perfect for empty nesters and mentions a nearby park that turns out to be two miles away. Another invents a renovated kitchen. The broker uses both, anonymized, in the next training session. Agents add a standard prompt that lists only file facts and forbids describing buyers, and the error rate in later samples drops.
Frequently asked questions
Is it legal for real estate agents to use AI?
Yes. The risks come from existing rules: fair housing[5], truthful advertising, privacy, review rules[6] and calling consent[7]. A written policy and human review manage most of them.
Do agents have to disclose AI use?
Rules vary, but clients expect it: 76% of Americans say being able to tell AI from human content is extremely or very important[2]. Disclose automated chat and staged images.
Can AI write my listing descriptions?
It can draft them, but the agent must check every fact and remove descriptions of who should live there before publishing.
Are AI voice calls allowed?
The FCC treats AI-generated voices as artificial under the TCPA[7], so telemarketing calls with them need prior express written consent.
References
National Association of REALTORS® (2026). REALTORS® Adopt Technology to Save Time and Improve the Client Experience, NAR Report Finds (2026 REALTORS® Technology Report) (opens in a new tab) ↑
Pew Research Center (2025). How Americans View AI and Its Impact on Human Abilities, Society (opens in a new tab) ↑
Gallup (2025). AI Use at Work Has Nearly Doubled in Two Years (opens in a new tab) ↑
Zillow Research (2026). Sellers: Results from the Zillow Consumer Housing Trends Report 2026 (opens in a new tab) ↑
U.S. Department of Housing and Urban Development (2026). Housing Discrimination Under the Fair Housing Act (opens in a new tab) ↑
Federal Trade Commission (2024). Federal Trade Commission Announces Final Rule Banning Fake Reviews and Testimonials (opens in a new tab) ↑
Federal Communications Commission (2024). FCC Makes AI-Generated Voices in Robocalls Illegal (opens in a new tab) ↑