AI Chatbots and Fair Housing: A Compliance Guide for Agents

AI assistants can answer buyer questions at 2 AM, which is a real advantage. They can also say things a licensed agent never should, such as describing who lives in a neighborhood or suggesting where a type of person “would fit in.” Under the Fair Housing Act, you are responsible for what appears under your name, including text an AI wrote. This guide explains the risk and the guardrails to demand from any tool. It is general information, not legal advice. Check with your broker and counsel.

What the Law Says, in Brief

The federal Fair Housing Act prohibits discrimination in housing based on protected characteristics: race, color, national origin, religion, sex, familial status and disability. Many states and cities add more. The prohibition covers advertising, statements and steering, whether done by a person or software. HUD published guidance in 2024 stating that the Act applies to housing advertising and tenant screening that use algorithms and AI. See HUD’s news release on the guidance and check HUD’s site for the latest, because federal guidance can change.

The principle is stable regardless of how guidance evolves: using AI does not remove your responsibility.

Where Chatbots Create Risk

1. Steering by neighborhood description

A buyer asks, “Is this a good neighborhood for families?” An ungrounded AI might answer with a description of the people who live there. Describing a community by its residents, rather than by verifiable features, can be steering or a discriminatory statement.

2. Answers from the open web

If a chatbot pulls from general internet text, it may repeat stereotypes, outdated crime or school claims or subjective opinions. You cannot supervise what you cannot predict.

3. Inconsistent treatment

If the assistant gives different information or different encouragement depending on how a person writes, that can create exposure. Consistent, factual answers reduce it.

4. Targeting and filtering

If you use AI to decide who gets an ad or a follow-up, be aware that targeting based on protected characteristics or proxies for them is risky. HUD’s advertising guidance focuses on this.

Guardrails to Require in Any Chatbot

1. Answers only from the listing facts you provide. Bedrooms, size, features, price, year built, HOA, what is included. No free-form commentary about people.
2. Says “I will check” when it does not know. It must not invent answers.
3. Refuses to describe who lives in an area. It should redirect to objective, official sources, such as a school district’s own lookup or the city’s official data.
4. No legal, financial or fair housing advice, and no promises on your behalf.
5. Full transcripts you can read. Audit them.
6. An approval mode. The option to review messages before they go out.
7. Clear disclosure that it is an assistant.

What a Safe Answer Looks Like

Buyer asks	Risky answer	Safer answer
“Is this a family neighborhood?”	Describes the kinds of people who live there	Shares listing facts and nearby amenities, and points to official school and city resources
“Are the schools good?”	Gives a subjective opinion	“School assignments depend on the district. Check the district’s own lookup. I can share the link.”
“What kind of people live around here?”	Answers at all	Declines and offers listing facts or a call with the agent
“Is it safe?”	Opinion	Points to official public data sources and offers a call

Prepare Your Listing Details

Most risky answers happen when the tool lacks a fact and fills the gap. Give it the facts. Use our listing FAQ guide and the free worksheet, which includes a reminder to link official sources instead of describing communities.

An Agent’s Fair Housing Checklist for AI

1. Read the vendor’s documentation on what the assistant can and cannot say.
2. Test it with hard questions: neighborhood, schools, safety, family status.
3. Review the first 20 transcripts, then weekly.
4. Turn on approval mode for outbound messages if available.
5. Keep your own copy of the transcripts.
6. Tell your broker you are using it and follow their policy.
7. Update the listing details when the assistant reveals gaps.

How ListingFlare Approaches It

The ListingFlare assistant is given the listing details you entered and is instructed not to make up details that are not in them, to say it needs to check when it does not know, and to say plainly that it is an AI assistant if someone sincerely asks. Its instructions do not include a specific fair housing rule or a filter for questions about neighborhoods or the people who live there, so test it with the questions above before you rely on it. You can read every conversation in your dashboard. AI Approval Mode lets you review the automatic reply to a new lead before it is sent; answers in the live chat on the listing page are not held for approval. No AI is perfect, so review conversations and keep your listing details accurate. Read how it works on the AI chat page, compare options in best real estate chatbots or test it yourself with the questions above.

For general fair housing resources, start with HUD’s Office of Fair Housing and Equal Opportunity.