Yes, it can draft them from facts you supply. Give it the verified features and the seller's priorities, ask for several versions, and delete every claim you cannot confirm. Then read each draft for wording about who a home suits or what the neighbors are like, because the Fair Housing Act covers advertisements however they were written.
What to give it
An assistant writes only as well as the sheet you hand it. Paste a short fact sheet from your own file: bedrooms, baths, square footage and its source, year built, updates with their years, lot size, parking, fees, and the two or three things the seller wants noticed. Ask for three versions of different lengths, one for the MLS remarks and one for a social post, and tell it to use no feature that is not on the sheet.
The tool does not know the house. It has not seen the kitchen, the street or the light at four in the afternoon, so photos and your own visit supply the sensory detail. Add that yourself in a final pass, which is also where the draft starts to sound like you.
What it gets wrong
Generated text can be confidently wrong. NIST defines confabulation as the production of confidently stated but erroneous or false content1. In a listing this shows up as a "gourmet kitchen" over a standard one, a "walk to downtown" with no distance behind it, a rounded-up square footage, or a feature from the house the model has seen most often. A buyer relies on the text, and so does an appraiser or an inspector reading your remarks.
Check each sentence against the fact sheet and against what you saw. A line that fits neither goes.
The fair housing read
We are not lawyers. Read the rule yourself or ask your counsel how it applies to you.
Section 3604(c) of the Fair Housing Act, 42 U.S.C. 3604(c), bars an advertisement that shows a preference or limitation based on a protected trait. An AI draft can slip into phrases that do exactly this, because it learned from decades of listings. Read for words about who a home suits and about who lives nearby. Replace them with facts about the property and distances.
| If the draft says | Write instead |
|---|---|
| Perfect for young professionals | Two blocks to the light rail station |
| Family-friendly street | Fenced yard, four bedrooms, cul-de-sac |
| Quiet, mature neighborhood | Tree-lined street, no through traffic |
| Near the best churches | 0.4 miles to the shopping center |
The right-hand column is our suggestion, and the checks on distances and counts are yours. Utah's consumer protection chapter adds that it is no defense that generative AI made a violative statement2. That is one state's rule on consumer protection, and it reflects how regulators look at an ad: they judge what you published and not the tool that wrote it. The tribunal in Moffatt v. Air Canada said it makes no difference whether information comes from a static page or a chatbot3.
A routine that takes ten minutes
Keep the fact sheet with the listing file. Generate the versions. Strike every claim not on the sheet, read the remainder aloud for any phrase about people, add one detail you saw yourself, and send the final to the seller for approval. Keep the sheet, because it shows where each claim came from if anyone asks later.
People also ask
Is AI virtual staging allowed in listings?
Whether and how you may use staged images depends on your MLS rules, your state and your code of ethics. The NAR code bars misleading images, so label staged photos clearly, show the original too and ask your broker or your MLS before you upload.
How can I use AI for real estate marketing?
Use it for first drafts: listing text, captions, neighborhood guides and email series. Add your own photos, numbers and voice. Describe the features of the home and the area, not the people in it, and make sure your firm name and licensure appear where required.
Can an AI agent answer my calls and leads?
An AI receptionist can answer while you are showing a home, say it is an AI assistant, capture the lead and book a call. Set it to hand off questions about neighborhoods and families. For outbound AI-voiced calls, an FCC ruling treats the voice as artificial, which calls for consent.
Which AI is best for real estate agents?
No single product wins, and we have not tested enough to rank them. Try the same fact sheet in two assistants and keep the one whose draft needs the fewest corrections. Judge each by whether you would stand behind its output in your own name.
Can I mention good schools in a listing?
We leave schools and safety out and state facts about the home and the area, such as distance to a named school. Words about who lives somewhere raise fair housing questions. Your broker's compliance contact has the final say.
Sources
- National Institute of Standards and Technology. Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, NIST AI 600-1. National Institute of Standards and Technology, 2024-07. nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf
- Utah Legislature. Utah Code 13-77-102 Liability for violation of consumer protection law. Utah State Legislature (le.utah.gov), 2026-05-06. le.utah.gov/xcode/Title13/Chapter77/C13-77-S102_2026050620260506.html
- Christopher C. Rivers, Tribunal Member. Moffatt v. Air Canada, 2024 BCCRT 149. Civil Resolution Tribunal of British Columbia, 2024-02-14. decisions.civilresolutionbc.ca/crt/crtd/en/525448/1/document.do