Fair Housing and AI-Written Listing Copy: The Phrases to Strike

AI writes listing copy in the voice of everything it has read, and a great deal of what it has read is decades of old real estate marketing — which is full of language that is now actionable. It will hand you Fair Housing problems cheerfully, fluently, at speed, and the copy will read beautifully. That combination is what makes this risk different from the ones agents are used to managing.

This post is the strike list, the reason behind it, and the one paragraph that stops most of it happening in the first place.

The rule underneath all of it

There is one test, and it replaces memorising a list:

If a sentence tells the reader what kind of person belongs here, it comes out.

Describe the property. Describe the amenities. Describe the location in terms of what is physically there. Say nothing about who should live in it — however well-meant, however obviously true, however much the seller asked you to.

Every phrase below fails that one test. That is why they are on the list.

The strike list

Family status and household composition

  • “Perfect for a young family”
  • “Great starter home for newlyweds”
  • “Ideal for empty nesters”
  • “Bachelor pad”
  • “Family-friendly neighborhood”
  • “Great for a growing family”
  • “Mother-in-law suite” — describe it as a second living area or accessory dwelling instead

Familial status is one of the seven federally protected classes. It covers households with children under 18, pregnant people, and anyone in the process of securing custody. “Perfect for a young family” and “ideal for empty nesters” are the same violation pointing in opposite directions.

One thing worth saying about that list of seven, because it trips agents up: since 2021 HUD has enforced the Act’s prohibition on sex discrimination as covering sexual orientation and gender identity. So “the seven classes” is accurate about the statute and incomplete about how it is enforced. Treat both as off limits in copy.

Race, colour, religion and national origin

  • “Walking distance to [named church, synagogue, mosque, temple]”
  • “Christian community”
  • “Exclusive neighborhood”
  • “Traditional neighborhood” — depends entirely on context, and the context is usually the problem
  • “Up-and-coming area”
  • “Changing neighborhood”
  • Any description of the demographic makeup of an area, in any direction, for any reason

“Exclusive” and “up-and-coming” are the two that surprise people. Both have long histories as coded references to who lives somewhere. If you want to say the house is expensive, say the price. If you want to say the area is being redeveloped, name the specific development.

Disability

  • “Perfect for active seniors”
  • “Not suitable for those with mobility issues”
  • “Must be able to manage stairs”
  • “Great for someone who can handle a project” — when describing accessibility rather than renovation

Two corrections here, because both get repeated wrongly in training rooms. “Handicap accessible” is not a violation in itself. The Fair Housing Act uses “handicap” as its statutory term and South Carolina’s law uses “physical or mental handicaps,” and describing an accessible feature is describing the property — which is exactly what you are supposed to do. Two caveats. HUD’s old advertising regulation was rescinded, so the word lists circulating in training rooms are industry guidance rather than a safe harbour. And HUD itself now writes “disability,” so “accessible” is simply the better word even where “handicap” is not actionable. The real risk with accessibility language is different anyway: it is claiming a feature you have not verified. Write “wheelchair accessible” and you are warranting a set of physical specifications. Measure it or do not say it.

The second correction: age is not a federal protected class in housing. People assume it is, because it is protected in employment. Where age-related language gets you in trouble is when it functions as a proxy for familial status — “empty nesters,” “no kids,” “adult community” — or when it misrepresents a community’s legal status. Qualifying 55-and-over communities are lawful under the Housing for Older Persons Act, but only if the community actually meets HOPA’s requirements. Do not describe a community as 55+ because the current residents happen to be.

Schools, safety, and crime

  • “Good schools”
  • “Top-rated school district”
  • “Safe neighborhood”
  • “Low crime area”
  • “Quiet, safe street”

These are the four I strike most often, and the ones agents argue with me about most. Two reasons they come out. First, school quality and crime rate have well-documented correlations with race, which makes them classic proxies — this is the steering exposure, and it is the one that produces complaints. Second, you cannot verify them. You are not the authority on whether a neighborhood is safe, and if the buyer’s experience differs from your marketing you have handed them a grievance with your name on it.

What to write instead: name the attendance zone as a fact and point the buyer at the district’s own data. “Zoned for [named school] — verify attendance zones and school performance directly with Charleston County School District.” You have given them the useful information and made no claim.

Free guide

The longer version of this is in The Real Estate Agent’s Guide to Claude & Cowork — 26 pages, written for South Carolina agents, with twenty-six prompts you can copy and use. No charge. Download it here.

Why AI produces this without being asked

It is worth understanding the mechanism, because it tells you how to stop it.

These models learned to write by reading enormous quantities of text, including a great deal of real estate marketing written before and during an era when this language was normal. When you ask for “warm, inviting listing copy for a three-bedroom bungalow,” the model reaches for the patterns that appear most often in warm, inviting listing copy. “Perfect for a young family” is one of the most common sentences in the entire genre. It is not being careless. It is doing exactly what you asked, using the material it has.

Which means the fix is not vigilance after the fact. It is constraint before the fact.

The paragraph that goes in every marketing prompt

Add this to every prompt where AI writes anything a consumer will see — listing remarks, social captions, email blasts, flyers, video scripts:

Follow Fair Housing rules strictly. Describe the property, never the people. No references to family status, age, race, colour, religion, national origin, disability, sex, sexual orientation, gender identity, or household makeup. No school quality claims and no safety or crime claims. Do not describe who the property would suit. If you are unsure whether a phrase is acceptable, leave it out and flag it to me.

That last sentence does real work. It converts the model from a writer that guesses into one that asks, and the flags it raises are a decent training exercise in themselves.

What this does not do

It does not make you compliant. It substantially reduces how often the problem is generated — it does not remove your obligation to read every word before it publishes.

It also does not cover steering, which is a conversation problem rather than a copy problem, or the separate question of what your brokerage requires. And the seven federal and state classes are not the whole list here. Several Lowcountry municipalities — the City of Charleston and Mount Pleasant among them — have adopted local ordinances adding sexual orientation and gender identity protections in housing, which neither South Carolina law nor the federal statute names explicitly. Published summaries of which municipality covers what do not fully agree with each other, so do not take mine or anyone else’s word for it: check the current ordinance with the municipality your listing sits in.

And nothing here is legal advice. Your broker-in-charge and your brokerage’s counsel are the authority on your obligations.

The rest of it

My free guide for agents has a full chapter on this — the strike list above, the prompt paragraph, the data-handling lines, and what never goes into an AI tool. 26 pages, no charge.

Download the guide.


Barry Cunningham runs Lowcountry AI Workshop and holds a South Carolina real estate license. South Carolina’s Fair Housing Law is enforced by the South Carolina Human Affairs Commission and covers race, colour, religion, sex, national origin, physical or mental handicaps, and familial status. This post is education, not legal advice.