Using AI in Real Estate Marketing: Rules and Risks for Teams and Brokerages

Quick answer: No law bans real estate agents from using AI in marketing. The rules that already apply, like fair housing, advertising, texting consent and license law, still apply when AI does the work, and the licensee and broker stay responsible for anything AI produces. A few newer rules target AI directly, mostly around digitally altered listing photos and bots that must say they are bots.

This is general information, not legal advice. AI rules are changing quickly and vary by state and MLS. Check with your broker of record, your MLS, your state commission or an attorney before adopting AI tools or policies.

On this page: Why it matters · Responsibility · Photos · Listing copy · Ad targeting · Texting assistants · Reviews · Copyright · Data and fraud · AI policy

Why AI rules matter now

AI is now part of everyday real estate marketing. In NAR’s 2026 technology report, nearly half of Realtors said they use AI daily or weekly, and 75% of those using it write listing descriptions with it. Social posts and email follow-up are close behind.

That speed is useful. It also means one unchecked AI draft or photo can reach the public in minutes, under your brokerage’s name.

The broker is still responsible

Regulators are clear on one point: the licensee and broker own the output, not the software vendor.

• South Carolina put it in statute: a licensee is responsible for all work product produced with the help of artificial intelligence.
• California’s Department of Real Estate says a broker’s duty to supervise extends to AI-powered software, and to treat AI like an unlicensed assistant.
• North Carolina’s commission says chatbots are fine for FAQs, scheduling and intake, but should not give legal, financial or brokerage-specific advice, and consumers should know they are dealing with an automated system.

A useful rule of thumb: an AI tool can do what an unlicensed assistant can do in your state, and nothing more.

AI-edited photos and virtual staging

This is where the most new rules have landed.

• California (since January 1, 2026): if an ad for the sale of real property uses a digitally altered image, it must say so and include a link, web address or QR code to the original, unaltered image. Routine edits like lighting, color correction, cropping, straightening and exposure are excluded.
• Wisconsin (from January 1, 2027): ads must disclose when images have been altered, including with AI, to add, remove or change elements in a way that creates a false or misleading impression.
• New York has bills pending that would require similar disclosures. They are proposals, not law, as of this writing.

MLS rules are tightening too. California’s CRMLS requires altered images to be labeled “altered,” “digitally altered” or “AI altered,” with the original shown right before or after. Stellar MLS requires “Virtually staged” captions and bans virtual staging on exteriors. ARMLS requires a “Digitally Altered” watermark paired with the original. Rules differ by MLS and several changed in 2026, so check yours.

NAR’s Code of Ethics already covers this. Article 12 requires a true picture in advertising, and Standard of Practice 12-10 bars manipulating listing content in a way that produces a deceptive or misleading result.

The safe habit: label every staged or altered photo, keep the original, and never use AI to hide defects or add features the home does not have.

AI-written listing copy

AI drafts listing descriptions fast, and two problems show up again and again:

• Made-up facts. AI tools can invent features, upgrades, square footage or school details. Publishing them is ordinary misrepresentation, whoever wrote the words.
• Fair housing language. AI can repeat phrases that describe people instead of the property, like “perfect for young families” or “great for retirees.” The Fair Housing Act applies no matter who or what wrote the ad. See Fair Housing Rules for Real Estate Social Media and Ads.

Have an agent check every AI draft against the MLS listing, tax records and the seller before it is published.

AI ad targeting and fair housing

Ad platforms use AI to decide who sees each ad, which is why housing ads carry special rules. Meta requires the Housing special ad category, with no targeting by age, gender or ZIP code, and it changed its ad delivery system for housing ads after a 2022 settlement with the Justice Department.

HUD withdrew its 2024 guidance on AI and digital ad platforms in 2025, but the Fair Housing Act and its advertising rules did not change.

New state laws on automated decisions are coming. Colorado’s new law, effective January 1, 2027, covers AI used in “consequential decisions,” including leasing or buying a home, and California’s privacy regulations on automated decision-making start the same day. Marketing copy and lead routing alone are probably not covered, but tenant screening or eligibility tools could be. Watch both if you use AI to screen applicants.

AI texting and chat assistants

AI assistants that text or chat with leads can respond in seconds, day or night. They also fall under texting and consent rules. The basics:

• Consent first. Text only people who gave consent, and record where and when they gave it.
• Say it is automated. Maine requires chatbots that could be mistaken for a person to disclose they are not human. Utah requires disclosure in some situations, and California bars bots from hiding what they are to drive a sale. Disclosing up front is good practice everywhere.
• Honor opt-outs. Since April 2025, people can opt out by any reasonable means, and you must honor it within 10 business days. Carrier rules require honoring STOP.
• Respect quiet hours. Federal rules bar telemarketing before 8 a.m. or after 9 p.m. local time. Florida, Oklahoma and Maryland are stricter, with an 8 p.m. cutoff and limits on attempts per day.
• Register your texting. Business texting through most platforms requires 10DLC registration, and unregistered traffic gets blocked.
• Hand off to a person. Send the lead to a licensed agent for pricing opinions, negotiation, contract terms, agency questions, complaints, or when the person asks for a human.
• AI voice is different. A 2024 FCC ruling treats AI-generated and cloned voices as artificial voices, so AI calls need prior consent, and written consent for marketing calls.

For how an AI assistant fits into a team’s lead system, see Lead Generation Systems for Real Estate Teams.

Reviews, testimonials and AI claims

The FTC’s rule on consumer reviews and testimonials, in effect since October 2024, bans fake reviews, including AI-generated reviews and testimonials from people who never worked with you. Violations can carry civil penalties of more than $53,000 each, and the FTC sent warning letters under the rule in December 2025.

Be careful with claims about your own AI, too. The FTC has brought cases over exaggerated “AI-powered” claims. “Our AI sells homes faster” needs proof.

Who owns AI content

• AI-only content generally is not copyrightable. The U.S. Copyright Office says prompts alone do not give enough human control, and the courts agree that an author must be human. The Supreme Court declined to hear the leading case in March 2026. Human writing, editing, selection and arrangement can still be protected.
• MLS data and photos have their own limits. Some MLSs bar feeding listing data into AI tools without permission, and photographers often keep the copyright in listing photos. Do not run another broker’s listing photos or MLS data through an AI tool your license does not allow.

Client data and AI-powered wire fraud

• Keep client data out of consumer AI tools. NAR’s guidance on AI policies says not to enter confidential or personal information into AI tools, and to review all output for accuracy and bias. Use business-grade tools with terms that bar training on your data.
• Fraud is getting more convincing. The FBI reported real estate fraud losses of more than $275 million in 2025, up from about $174 million in 2024, and warns that criminals use AI to clone voices and create fake video. Verify wire instructions by phone using a known number, and consider a verification code word for any urgent money request.

Safe uses vs risky uses

Use	Lower risk when	Higher risk when
Listing description drafts	An agent checks facts and fair housing language	Published unedited, with invented features or phrases about people
Photo cleanup	Only lighting, color, crop or straightening	Removing power lines, cracks or neighbors, or adding features
Virtual staging	Labeled per MLS and state law, with the original available	Unlabeled, on exteriors where banned, or hiding defects
AI texting assistant	Consented leads, clear disclosure, opt-outs honored, fast handoff to an agent	Texting purchased lists, posing as a person, answering pricing or contract questions
AI voice calls	Documented prior consent	Cold calls with an AI or cloned voice
Social posts and emails	Reviewed by a person for accuracy and brand	Fake testimonials or AI-written reviews
Ad targeting	Housing ad category and broad geography	Demographic targeting or AI-picked audiences nobody reviews
Market summaries	An agent verifies the data and signs off	Presented as a licensed opinion without review
Client data	Business tools with no-training terms, minimal personal data	Pasting financials, IDs or contracts into free AI tools
Claims about your AI	Accurate descriptions	“Our AI guarantees a faster sale”

A brokerage AI policy

NAR recommends that brokerages adopt a written AI policy. A practical one covers:

1. An owner: the broker of record or a named manager approves AI tools and keeps a list of approved ones.
2. Human review: a licensee reviews every AI output before it is published or sent.
3. Images: label altered and staged photos, keep originals, and follow your MLS and state rules.
4. Assistant disclosure: AI assistants say they are automated at the start of each conversation.
5. Consent and opt-outs: record consent, honor opt-outs quickly, respect the strictest quiet hours that apply, and keep texting registration current.
6. Handoff triggers: pricing, negotiation, contracts, agency, fair housing sensitive questions, complaints, and any request for a person go to a licensed agent.
7. Ad targeting: use housing ad categories and review AI-chosen audiences.
8. Reviews and claims: no AI-written or fake reviews, and no inflated AI claims.
9. Data rules: no client personal information in unapproved tools, and check vendor terms on training, retention and security.
10. Fraud controls: verify wire instructions by phone and treat urgent voice or video requests with suspicion.
11. Records: keep published ads, consent and opt-out logs, chat transcripts and handoff records for your state’s retention period.
12. Training and review: train agents yearly, spot-check AI output quarterly, and watch for new state and MLS rules.

A marketing leader, full-time or fractional, usually owns this policy with the broker. A social media manager or marketing generalist can run the day-to-day checks.

Key takeaways

• The rules that already apply to marketing still apply when AI does the work.
• The licensee and broker are responsible for AI output, not the vendor.
• Label altered and staged photos and keep the originals. California requires it now, and Wisconsin from 2027.
• Review every AI listing draft for facts and fair housing language.
• AI texting assistants need consent, disclosure, opt-outs, quiet hours and a fast handoff to a licensed agent.
• Put it in a written AI policy and train on it every year.

Sources

Show all 14 sources

• National Association of Realtors, 2026 technology report and Crafting an AI use policy
• South Carolina Real Estate Commission, AI use guidance
• California Department of Real Estate, Advisories
• NC Real Estate Commission, Using chatbots and automated responses
• California Legislature, AB 723 (digitally altered images)
• Wisconsin, 2025 Act 69
• CRMLS, Digitally altered image guidance, Stellar MLS, Photo rules and ARMLS, Digitally altered media
• National Association of Realtors, 2026 Code of Ethics
• Colorado General Assembly, SB 26-189 and California Privacy Protection Agency, ADMT regulations
• Maine, Chatbot disclosure law (P.L. 2025, c. 294) and California, Bot disclosure law
• Federal Communications Commission, Ruling on AI-generated voices (2024) and Consent revocation rule
• Federal Trade Commission, Rule banning fake reviews and testimonials and December 2025 warning letters
• U.S. Copyright Office, Copyright and artificial intelligence
• FBI Internet Crime Complaint Center, 2025 report and Generative AI fraud warning

Frequently asked questions

Do I have to disclose virtual staging?

Almost always. Most MLSs require virtually staged or altered photos to be labeled, and some require the original photo too. California law has required a disclosure and a link to the original since January 1, 2026, and Wisconsin requires disclosure starting January 1, 2027. Routine edits like lighting, color and cropping generally do not count.

Can our AI assistant text leads?

Yes, with guardrails. Text only people who gave consent, say it is an automated assistant, honor STOP and other opt-outs, respect quiet hours, register your texting program, and hand the lead to a licensed agent for anything that needs a license. AI voice calls need prior consent under a 2024 FCC ruling.

Is AI-written listing copy a fair housing risk?

It can be. AI tools can repeat steering phrases like perfect for young families, or invent features and square footage. Fair housing and advertising rules hold the licensee and broker responsible, not the software, so an agent should review every AI draft before it is published.

Who owns AI-generated marketing content?

Content made purely by AI generally cannot be copyrighted in the US. Human writing, editing, selection and arrangement can be protected. Check the AI tool's terms, the photographer's license and your MLS data license before reusing content.