Pre-check every housing ad before it publishes

The failure mode for any automated fair-housing ad check is not missing things. It is training the leasing team to click past it. A tool that flags every sentence that happens to contain the word "family" ends up ignored on the "adults only" line that actually carries HUD risk. A pre-publish check has to earn its place on every listing: deliberate about what it blocks on, what it warns about, and what it refuses to judge. The Fair Housing Act's advertising prohibition at 42 U.S.C. 3604(c), read with HUD's discriminatory-advertising rule at 24 CFR 100.75, is useful here as an ordering of attention, because the rule already separates wording that is categorically an indication of preference from wording that is merely fixable by describing the property more neutrally.

Block on what is categorically an indication in the wording

Some constructions are facts about the paste, not opinions about the unit. A line that names a group in a prefer-or-exclude frame ("no [group]," "ideal for [group]," "restricted to," "won't rent to") hits 24 CFR 100.75 and 3604(c) directly, because the protected bases in 42 U.S.C. 3602 are exactly what such a line points at. A "no children / adults only" line or a couple-coded "perfect for a young professional" is a familial-status indication. A "must be able to climb the stairs" screen, or a flat "no pets" with no assistance-animal exception, is a disability indication. Block on these before the listing goes live, because these are the findings that become complaints, and the only honest uncertainty is how a given agency weighs them - not whether the pattern is present in the text.

Warn on what is fixable by rephrasing to the property

The middle tier is where a good tool earns its keep. A coded descriptor - "exclusive community," "safe and desirable area," "walking distance" phrased as a mobility test - is usually fixable by replacing a subjective, demographic characterization with an objective, verifiable fact about the property and its physical surroundings. Warn, quote the sentence, and let a human rephrase. This tier is valuable because the fix is usually one line and the failure is genuinely easy to make when describing a neighborhood from memory. The steering prohibition at 24 CFR 100.70 is the anchor for coded steering wording, and it turns on whether the sentence guides a reader toward or away from a place because of who they are.

Leave alone what cannot be decided from the text

Whether an applicant was actually turned away, whether the ad was targeted to or away from a protected group on a platform, whether a rendered Equal Housing Opportunity logo actually appeared in the image, and whether a given occupancy rule is a neutral code limit or a family screen dressed up - those live outside the pasted copy. A fair-housing checker that pretends otherwise produces confident noise. The honest move is to say plainly: I read your wording; I cannot see your applicants, your platform, or your pixels. A tool that scores an "approval probability" is lying about what a text scan can do, and the enforcement discretion that the FHA's policy statement hands to HUD, DOJ, and state and local agencies is not something a wording check can predict.

Practice logic beats a banned-word list

The naive approach is a list of forbidden words: block "children," block "pets," block "Christian." It fails twice. First, it blocks lawful uses - "two-bedroom unit, equal-opportunity occupancy up to four per local code, service and assistance animals welcome, Equal Housing Opportunity" is a model listing that a word list may still flag for "children" or "animals." Second, the rule is not a word list; 24 CFR 100.75 turns on whether an ordinary reader would see an indication of preference, and the same phrase can be a violation or perfectly fine depending on the frame around it. Whether a phrase reads as a preference in context is a judgment; whether it is present in the copy is deterministic to check, and conflating the two loses trust fast.

The tiers shift with channel and ad type

The FHA reaches advertising in any medium - the 3604(c) text covers making, printing, or publishing any notice, statement, or advertisement - so a yard sign, an MLS remark, a social post, and a listing-site paragraph all carry the same duty. What changes is the risk shape: a short yard-sign line has fewer words to hide behind, an MLS description can quietly accumulate coded neighborhood characterizations, and a paid social ad can be targeted in a way that compounds a borderline sentence. A pre-check that lets you say the ad type (rental listing, for-sale listing, agent or brokerage ad, social or MLS post) and the channel, and escalates accordingly, is more useful than one that treats every paste the same.

Keep the finding, not just the verdict

What teams actually want from a fair-housing check is not a raw score; it is a trustworthy ordering they will follow when a listing is due out today. So a finding that names the exact phrase and the rule it maps to can be re-checked next quarter when a similar ad is drafted; a bare pass/fail cannot. Run it on a fixed cadence - one scan per listing draft, before it publishes, never after a complaint lands - and grade the habit, not the single ad. A green score on one paste means only that the particular sentences you pasted did not trip a known pattern; the safe program is the one where the risky wording never leaves the drafting window.

Where the pre-check stops

The pre-check reads the wording you paste. It does not read your tenant-selection criteria, your platform's audience settings, your rendered images, or your jurisdiction's full class list. Beyond the federal seven, HUD reads "sex" to include sexual orientation and gender identity, and many states and localities add source of income, age, marital status, military status, and more - all surfaced as advisories because the applicable list varies, never quoted as if universal. It is a shape-of-attention decision-support layer for a team that does not have the FHA and 24 CFR 100.75 at their fingertips the moment a listing is being drafted. When the tool is quiet, that is the moment to ask what it cannot see, not the moment to stop looking.

Heuristic pre-check only - not a fair-housing certification or legal advice. The federal basis is the Fair Housing Act (42 U.S.C. 3604(c)) with HUD's discriminatory-advertising rule (24 CFR 100.75). 16 CFR Part 310 is the Telemarketing Sales Rule, not the fair-housing standard. Sexual orientation and gender identity fall under HUD's reading of "sex," and state and local law may add protected classes - advisories only, not quoted. No intent, platform-targeting, rendered-logo, occupancy-code-interpretation, or enforcement verification is performed by any wording-only checker, including the one that links to this article. For a binding view on a specific advertisement, consult the Fair Housing Act (see the requirements-language discipline in RFC 2119 for how these rules are worded) and a qualified professional.