AI in real estate listing marketing: a checklist for brokerages on copy, photos and fair housing

Listing marketing is where AI arrived first in residential real estate. It is cheap, it is fast and the output looks good. An agent can turn a page of notes into polished listing copy in seconds, or turn an empty living room into a furnished one without hiring a stager.

That speed creates a new kind of risk for the brokerage. The copy, the photos and the ads all carry the brokerage’s name and licence, and regulators in both countries have been clear that "the AI did it" is not a defence. This checklist sets out what has changed, what the evidence shows and what a brokerage should have in place before its agents publish the next AI-assisted listing.

Why AI marketing became a brokerage issue

Adoption has moved quickly. In NAR’s 2026 technology report, 23% of US agents said they use AI daily and another 25% weekly, while only 12% said they do not use it and have no plans to ¹. Among agents who use AI, 75% use it to write listing descriptions, 56% for social media posts and 52% for emails and follow-ups ¹. The survey was fielded in June 2026 ². Results on value are mixed: 55% of agents reported a positive effect on their business, up from 50% a year earlier, but 11% reported a negative one, up from 4% ².

Images are the second big use. Staging matters to buyers: in NAR’s 2025 staging survey, 83% of buyers’ agents said staging made it easier for buyers to picture the property as their future home ³. AI virtual staging, which adds furniture and decor to photos of empty rooms, promises that effect for a fraction of the cost. The same survey found the median cost was $1,500 when a staging service was used ³. The survey drew 1,266 responses from NAR members in February 2025 ³.

The problem is disclosure. A study by Coraly, a real estate data platform that has built a disclosure compliance workflow with San Diego MLS, reviewed just under 40,000 primary listing photos on Zillow, Redfin, Realtor.com and Homes.com in the first quarter. It found 10.8% showed signs of digital alteration, and more than 90% of the altered images had no visible disclosure ⁶. Sky replacement made up 69% of the alterations and virtual staging 10% ⁶. These are vendor figures, but they point at a real gap.

The rules that now apply

California made disclosure a legal duty. A new section of the state’s Business and Professions Code, added by Assembly Bill 723 and signed in October 2025, requires brokers and salespeople who use a digitally altered image in advertising for the sale of property to say the image was altered and to provide a link, URL or QR code to the original ⁴. On websites the licensee controls, the unaltered version must be posted too, or linked ⁴. The definition covers images changed with photo editing software or AI to add, remove or change elements such as furniture, flooring, paint colour, landscaping, views through windows and neighbouring properties. Routine edits such as lighting, cropping, colour correction and straightening are excluded if they do not change how the property is represented ⁴.

California’s regulator spelled out the wider duties. A March 2026 advisory from the Department of Real Estate states that "all real estate advertising in California must be truthful, accurate, and not misleading" ⁵. It tells licensees to check AI-drafted facts such as features, pricing and availability themselves, says supervision duties extend to the tools agents use, and recommends written AI policies, training and records of how AI output was reviewed ⁵. Responsibility, it says, sits with the licensee and the responsible broker, not the technology provider ⁵.

Canadian regulators apply existing advertising rules. The Saskatchewan Real Estate Commission’s spring 2026 newsletter warns that AI tools can "hallucinate", meaning they produce content that is inaccurate or incomplete, and that the registrant remains responsible for a listing description with errors ⁷. Virtually staged photos can breach its bylaw on misleading advertising if they mislead about the property’s current condition ⁷. It recommends a visible label such as "Virtually Staged" or "Edited with AI", and asks brokers to set AI policies and supervise their use ⁷. In British Columbia, BCFSA’s Artificial Intelligence Guideline, issued in February 2024, keeps licensees accountable for AI-assisted work and calls for checking accuracy, protecting confidential information, avoiding bias and disclosing altered advertising media ⁸. Other provinces’ false advertising rules raise the same issues even without AI-specific guidance.

Fair housing does not change because a machine wrote the ad

The US Fair Housing Act’s advertising rule makes it unlawful to publish an ad for a dwelling that indicates a preference or limitation based on a protected class, and it names words, phrases, photographs and the choice of media or locations as ways to break it ⁹. That covers AI-written copy and AI-chosen audiences in the same way it covers human ones.

The federal guidance has shifted, but the law has not. In 2024 HUD issued guidance on how the Act applies to ads delivered through digital platforms, including AI-driven targeting. HUD withdrew that guidance, along with seven other fair housing documents, with effect from September 2025, saying it should not be relied on as authoritative ¹⁰. The statute and the regulation still apply, and California’s advisory makes the same point at state level ⁵.

The risk is not only theoretical. A 2024 MIT audit of GPT-4 using 168,000 prompts about renting and buying in the ten largest majority-minority US cities found evidence of racial steering and of minority home seekers being steered towards neighbourhoods with lower opportunity scores ¹¹. That study looked at housing recommendations rather than listing copy, but it shows why AI output about neighbourhoods and "ideal buyers" needs a human check. On ad delivery, Meta built a system to narrow gaps in who sees housing ads under a settlement with the US Justice Department, and it stopped offering its Special Ad Audiences tool for housing ads. Court oversight was set to run until June 27, 2026 ¹².

Reviews and testimonials

AI can draft client testimonials as easily as listing copy. In the US, the FTC’s rule on fake reviews bans reviews and testimonials that misrepresent who wrote them or that come from people without real experience, explicitly including AI-generated ones, and allows the FTC to seek civil penalties from knowing violators ¹³. Agents can use AI to help clients put real feedback into words, but every testimonial should trace back to a real client who approved it.

The brokerage checklist

Listing copy

• Require agents to check every factual claim in AI-drafted copy against the listing file: rooms, square footage, upgrades, taxes, fees and availability.
• Ban AI-written descriptions of the "ideal buyer" or of neighbourhood demographics, and review copy for fair housing language before it goes live.
• Keep the brokerage name and other required identifiers in every ad, whatever tool produced it.

Images

• Label every virtually staged or materially edited photo on the image itself, with wording such as "Virtually Staged".
• Keep the original photo for every altered one, and in California link to it from the listing.
• Do not remove defects, power lines, neighbouring buildings or other permanent features, and do not change views from windows.
• Treat sky replacement and lawn greening as material edits unless a lawyer advises otherwise.

Targeting and ads

• Use the special housing ad categories on social platforms, and do not narrow housing ads by age, gender or postal code.
• Review where AI-optimised campaigns actually deliver ads, not only who they were aimed at.

Reviews

• Publish only testimonials from real clients, with their approval on file, and never generate reviews.

Governance

• Adopt a written AI marketing policy, train agents on it and name a broker or manager who owns it.
• Keep a simple record of which listings used AI copy or images and who reviewed them.
• Tell agents not to paste client personal information into public AI tools.

Practical next steps

1. Audit 20 current listings. Count how many use AI-written copy or altered photos and whether the photos are labelled. That gives you a baseline.
2. Fix images first. Unlabelled altered photos are the most visible gap and, in California, a legal violation.
3. Write a one-page policy. Cover copy checks, image labels, ad targeting, reviews and privacy, and get agents to acknowledge it.
4. Build review into the workflow. Add a check step in your listing launch process rather than relying on agents to remember.
5. Watch for new rules. Other US states and Canadian provinces are likely to follow California and Saskatchewan, and MLS rules are changing too.

Sources

1. National Association of REALTORS® (September 22, 2026). REALTORS® Adopt Technology to Save Time and Improve the Client Experience, NAR Report Finds. Member survey, sample size not stated in the release. Industry association.
2. Real Estate News, Dave Gallagher (September 22, 2026). Agents branching out from generative AI as adoption grows. Reports NAR’s 2026 survey, fielded June 2026.
3. National Association of REALTORS® (May 2025). 2025 Profile of Home Staging. Survey of 1,266 members, February 2025, margin of error 2.75 points. Industry association.
4. California Legislature (Chapter 497, Statutes of 2025, approved October 10, 2025). AB 723, Real estate: digitally altered images: disclosure. Adds Section 10140.8 to the Business and Professions Code.
5. California Department of Real Estate (March 17, 2026). Artificial Intelligence in California Real Estate: Opportunities, Risks, and Compliance Considerations for Licensees. Regulator advisory.
6. HousingWire, Jonathan Delozier (June 24, 2026). Most AI-altered listings go undisclosed, California law bans it. Reports a Coraly study of just under 40,000 listing images. Coraly offers a compliance workflow (vendor research).
7. Saskatchewan Real Estate Commission (Spring 2026). The Register: Use of Artificial Intelligence: Advertising. Regulator newsletter.
8. Pax Law, Samin Mortazavi (October 27, 2025). AI in Real Estate: Trends, Risks, and Best Practices in British Columbia. Law firm summary of BCFSA’s Artificial Intelligence Guideline (February 2024).
9. US Code of Federal Regulations, via Cornell Legal Information Institute. 24 CFR § 100.75, Discriminatory advertisements, statements and notices.
10. Orrick, Herrington & Sutcliffe, via JD Supra (April 13, 2026). HUD withdraws eight fair housing guidance documents as part of deregulatory review. Law firm summary of HUD’s April 6, 2026 Federal Register notice.
11. Liu, So, Hosoi and D’Ignazio, MIT (October 29, 2024). Racial Steering by Large Language Models: A Prospective Audit of GPT-4 on Housing Recommendations. EAAMO ’24, ACM. 168,000 prompts across ten US cities.
12. HousingWire, Flávia Furlan Nunes (January 10, 2023). Meta rolls out new system to address housing ad discrimination.
13. US Federal Trade Commission (August 14, 2024). Federal Trade Commission Announces Final Rule Banning Fake Reviews and Testimonials.