Disclosing an AI written listing description

Is there actually a rule requiring AI disclosure in a listing description?

Not one found in NAR's own governing documents, in the FTC's advertising guidance, or in either of the two MLS rulebooks reviewed directly for this guide. The claim that Article 12 of the Code of Ethics picked up new AI specific language for 2026 circulates in real estate marketing content, but the text of Standard of Practice 12-10 itself, checked against the current Code of Ethics, still reads as it did after its 2018 amendment: it prohibits "manipulating (e.g., presenting content developed by others) listing and other content in any way that produces a deceptive or misleading result." That sentence does not mention artificial intelligence, does not distinguish AI generated text from human written text, and was written years before generative AI tools existed in their current form. It has proven broad enough on its own to cover the new tools without needing a rewrite.

This matters because it changes the actual compliance question. The rule is not "did a machine write this," it is "is this true and not misleading," which is exactly the same test that applied to a paragraph a human copywriter wrote before any AI tool existed. An accurate AI drafted description that a licensed agent reviewed and approved raises no more of an ethics question under 12-10 than an accurate human drafted one does.

What is the difference between an AI written photo and an AI written description, legally?

A real one, and conflating them is the single most common mistake in coverage of this topic. California's AB 723, in force since 1 January 2026 and covered fully in California's AI photo disclosure law for real estate listings, puts a statutory disclosure duty specifically on digitally altered images, requiring a conspicuous statement and a link to the unaltered original. That statute is about pictures. It says nothing about text, and no comparable state statute requiring disclosure of AI generated listing prose was found for this guide. The MLS rules that require a virtual staging label, covered in disclosing a virtually staged photograph on the MLS, are also photograph specific by their own terms.

Text has no equivalent structure anywhere in the sources checked. There is no MLS rule requiring a "written with AI assistance" tag on public remarks, and no state statute doing the same for real estate advertising copy specifically. The asymmetry is not an oversight so much as a difference in what regulators have identified as the risk: an altered photograph can misrepresent what a buyer will physically see when they walk in, while a well drafted sentence, whoever or whatever produced its first draft, is judged on whether the claims in it are true.

Content type	Disclosure required?	Governing rule
Digitally altered listing photograph, California	Yes	AB 723, Business and Professions Code 10140.8
Virtually staged photograph, most MLSs	Yes, MLS specific label or watermark	CRMLS Rule 11.5.2, ARMLS watermark policy
AI written listing description, any state found	No disclosure duty identified	General deception standards apply instead
Any listing content, REALTOR members	No AI specific rule	NAR Standard of Practice 12-10, unchanged since 2018

Does the FTC treat AI written advertising copy differently from human written copy?

Not on the point that matters for a listing description. Section 5 of the FTC Act and the Commission's Policy Statement on Deception, dated 14 October 1983, judge an advertisement by whether a representation is likely to mislead a consumer acting reasonably under the circumstances, and that test runs on the content and the effect of the claim rather than on who or what typed it. A false square footage figure is a false square footage figure whether a person typed it from memory or an AI tool generated it from a prompt the agent supplied and then approved without checking. The FTC's endorsement and testimonial guidance addresses disclosure obligations around paid endorsements and material connections, a different scenario from a brokerage's own listing copy, and no FTC guidance found for this guide creates a distinct disclosure duty for AI authored advertising text as such.

The practical upshot is that the liability an agent already carries for an inaccurate description does not change based on drafting method. What changes is the temptation to trust an AI generated sentence without the same scrutiny a self written one would get, and that habit is the actual risk, not a missing disclosure line.

Is there a reason to care about AI authorship even without a disclosure rule?

Yes, and it runs through copyright rather than through advertising law. The U.S. Copyright Office's registration guidance, published in the Federal Register on 16 March 2023, states that "if a work's traditional elements of authorship were produced by a machine, the work lacks human authorship and the Office will not register it," and that applicants have a duty to disclose AI generated content when registering a work for copyright. A human can still hold copyright over material they selected, arranged or substantially rewrote from AI output, but purely AI generated prose, an unedited listing description generated end to end by a tool, likely is not protectable by copyright at all under that standard.

That has a concrete consequence for an agent who cares about a competitor copying their listing language. How to write an MLS listing description already covers CRMLS Rule 12.5's requirement that a subsequent agent get prior written authorisation before copying remarks, which is an MLS contract obligation independent of copyright. But the copyright backstop behind that MLS rule, the one that lets a photographer or a writer sue over unauthorised use of their work under 17 U.S.C. 504, only protects work with a human author. An unedited AI draft sitting in public remarks has weaker copyright standing than the same paragraph after a human materially rewrites it, which is a reason to edit AI output substantively even where no rule requires disclosing that AI touched it at all.

Should a brokerage set its own AI disclosure policy anyway?

Absence of a legal requirement is not the same as absence of a reason. A brokerage's errors and omissions carrier may ask, as part of underwriting, how listing content is produced and reviewed, and an internal policy that requires a licensed human to read and approve every AI drafted description before publication is a defensible answer regardless of what the law separately requires. That review step is also where the accuracy problem actually gets caught, since an AI tool has no way to verify that a stated lot size, a school assignment or a claimed renovation year is correct against the actual property, and a fair housing read of the language still has to happen whether a person or a tool produced the first draft.

Some brokerages choose to disclose AI assistance to sellers as a matter of client relations rather than compliance, on the reasoning that a seller paying for marketing services has a reasonable expectation of knowing how those services are produced. That is a business choice about client trust, not a response to a rule, and it can be made independently of anything covered above. Treating the two questions as separate, what the law requires and what a brokerage decides to promise its clients, keeps a marketing decision from being misdescribed to an agent as a compliance one.

Frequently asked questions

Do I have to tell a seller their listing description was written by AI?

No law or MLS rule reviewed for this guide requires it. Some brokerages set their own internal policy requiring disclosure to clients or requiring agent review before publication, which is a business decision rather than a legal one, and it is worth checking your own brokerage's policy since it can be stricter than what the law demands.

Does NAR's 2026 Code of Ethics have a new AI disclosure rule?

No. Standard of Practice 12-10, the provision most relevant to manipulated or misleading content, was adopted in 2007 and amended in 2018, and that text carried into the 2026 Code of Ethics unchanged. Content claiming a new AI specific clause for 2026 is not supported by the text of the Code itself.

If AI wrote my description and it turns out wrong, who is liable?

The agent and brokerage that published it, on the same basis as any other inaccurate advertising claim. Drafting method does not change the underlying duty to publish accurate information, and a claim that "the AI wrote it" is not a recognized defense against a false advertising or misrepresentation claim in any source checked for this guide.

Can I copyright a listing description an AI wrote for me?

Only the parts you substantially authored yourself. The U.S. Copyright Office's position is that purely machine generated text lacks the human authorship copyright requires, so an unedited AI draft likely has no copyright protection at all, while a version you meaningfully rewrote or arranged can be protected for your own contribution.

Sources

Last verified: 2026-09-02