Fair Housing tips for AI conversations
An AI assistant speaks on your behalf, so its messages carry the same Fair Housing obligations as your own. This is general practical guidance, not legal advice — confirm your process with your broker and counsel.
Never filter or comment on protected classes
US federal law protects race, colour, national origin, religion, sex (including sexual orientation and gender identity), familial status, and disability. Many states and cities add source of income, age, marital status, and more.
- Do not ask about, record, or act on any protected characteristic.
- Do not describe a neighbourhood by who lives there. Describe amenities, transit, and facts instead.
- Do not answer "is this a good area for families like mine?" with a recommendation — provide neutral data sources and let the buyer decide.
Avoid steering in listing suggestions
Recommend properties on stated criteria only: budget, size, timeline, and required features. Never vary the set of listings shown based on an inferred characteristic, and keep the same qualifying questions for every enquirer.
Keep screening consistent
- Apply identical criteria and identical wording to every applicant.
- State screening criteria up front in the knowledge base so nobody receives a different version.
- Accommodate disability-related requests and assistance animals per policy — escalate these to a human every time.
What to escalate to a human
- Any question that touches a protected characteristic.
- Reasonable accommodation or modification requests.
- Complaints, disputes, or anything that sounds like a legal claim.
Transcripts are your evidence that every enquirer was treated the same way. Keep them for at least as long as your brokerage's record retention policy requires.
Frequently asked
Is an AI assistant allowed under Fair Housing rules?
Yes — the obligations attach to the messages, not the sender. Using an AI is fine provided it asks the same questions of everyone, never filters on protected characteristics, and escalates sensitive topics to a licensed human.
Who is responsible if the AI says something non-compliant?
You are, as the licensed professional or brokerage. That is why supervised rollout, escalation rules, and regular transcript review are part of every guide here.
This guide is general operational guidance maintained by the Lytle.ai team and is not legal advice. Confirm your process with your broker and qualified counsel for your jurisdiction.