AI Tools for Real Estate Agents, With the Rules That Apply

Using AI in a real estate practice is allowed, and most agents already lean on it for listing copy, email drafts and follow-up.

Trouble starts when output goes out unchecked: a description that signals a preferred buyer, a staged photo presented as real, or an automated message sent to someone who never agreed to receive it.

Those three problems map to three rules that decide which tools you can use and how. This guide starts with the rules, sorts everyday tasks by risk, follows one listing from signing to closing, and then covers the tools and the questions to put to any vendor.

The rules below are general information drawn from agency guidance, MLS notices and law-firm summaries. They vary by state and MLS, so confirm the details with your broker and a real estate attorney before you rely on them.

Three Rules That Decide What You Can Use

Rule 1: Fair housing covers AI-written ads

HUD said in April 2024 that the Fair Housing Act applies to the advertising of housing, including when algorithms and AI help create or deliver the ads, as HUD's press release states. Its advertising guidance notes that discriminatory targeting can be deliberate or can result from how a platform delivers ads.

The safe assumption is that an AI draft is held to the same standard as copy you wrote yourself. Watch for phrases about who a home or neighborhood is "perfect for," and for references to churches, schools or demographics that imply who belongs.

NAR's broker guidance, as HousingWire summarizes it, warns that AI-generated content may be inaccurate, can create fair housing risk and must still meet the truthful-advertising duties in Articles 2 and 12 of the Code of Ethics.

Also Read: The Ethics and Risks of AI in the Workplace: What Every Business Needs to Know

Rule 2: Altered photos must be disclosed, and in California the original must be available

California's AB 723 took effect on January 1, 2026. It added section 10140.8 to the Business and Professions Code and requires a licensee, or someone acting for one, to disclose digitally altered images used to advertise real property for sale and to give access to the original, according to San Diego MLS's AB 723 FAQ.

That FAQ says images created or changed with AI, including AI virtual staging, count as altered. It also says labeling alone falls short, since buyers must be able to view the original image.

Because the rule sits inside the licensing law, a law firm briefing says a willful failure to disclose can be a criminal offense. MLSs have followed with matching rules, including CRMLS in its December 2025 broker report.

What counts as altered is a judgment call. San Diego MLS describes it as an edit that materially changes how the property appears, and it says that when in doubt, disclosure is strongly recommended.

CRMLS goes further for its members. A San Diego Association of Realtors notice says CRMLS Rule 11.5(c) does not allow digitally altered images other than virtual staging.

CRMLS also tells members who are unsure how much an image has changed to err on the side of caution and include the original.

One law firm's guide offers sample caption wording for a staged image:

This image has been digitally altered for marketing purposes. Buyer should review the property in person and verify all features and conditions.

Treat that as a model, as Barnes & Walker's guide presents it, and ask your broker which wording your MLS prefers.

Outside California, the same habit protects you. NAR says the true-picture standard in Article 12 applies to listing photos and advises labeling virtually staged images, as HousingWire reports, and many MLSs have their own staged-photo rules.

Rule 3: AI voices and automated messages need consent

In February 2024 the FCC ruled that calls using AI-generated or cloned voices are "artificial or prerecorded voice" calls under the Telephone Consumer Protection Act.

Callers generally need prior express consent, and prior express written consent for marketing calls, as a Wiley summary explains. Callers must also identify themselves, and marketing calls must offer an opt-out.

Later that month the FCC adopted rules saying consumers can revoke consent by any reasonable means. Texting words such as "stop," "quit," "end," "revoke," "opt out," "cancel" or "unsubscribe" counts, according to Kelley Drye's TCPA tracker.

For an agent, that means keeping a record of how each lead agreed to be contacted, and confirming that any AI follow-up tool stops when someone says stop.

A common approach is a checkbox on the web form that names the channels you will use, with the submission stored and time-stamped. Ask your broker what consent language and records your brokerage requires.

None of the three rules bans AI. They set the conditions under which AI can speak to the public on your behalf.

A Traffic-Light Map of Everyday AI Tasks

Sorting tasks by risk shows where AI can run freely and where it needs a checkpoint:

Green: low risk, with a quick review

• Email and message drafts: Edit for accuracy and tone before sending anything to a named client.
• Meeting and call summaries: Get consent before you record or transcribe a conversation.
• Checklists and showing-day plans: Generic planning documents carry little risk.

Yellow: needs a human check and a record

• Listing descriptions: Verify every feature against the property and screen the language for fair housing before it reaches the MLS.
• Virtual staging and photo edits: Label every altered image and keep the originals.
• Pricing and market summaries: Treat AI figures as drafts, and confirm comparable sales against MLS data.
• Housing ads on social platforms: HUD's digital advertising guidance applies, so use each platform's housing-ad settings and review the targeting.

Red: avoid, or get approval first

• AI calls and texts without recorded consent: The FCC ruling puts these under the TCPA.
• Screening buyers or tenants with an AI tool: HUD issued separate guidance on AI in tenant screening, so involve your broker first.
• Pasting client financials or access details into public AI tools: NAR's guidance warns agents to protect personal information, as HousingWire reports.

Also Read: How to Automate Repetitive Daily Tasks Using AI

One Listing, Start to Closing: Where AI Fits

Following a single sale shows where each rule appears:

1. Listing appointment: Use an assistant to organize your notes and draft a preparation outline, and keep the seller's financial details out of public tools.
2. Photos and staging: If you virtually stage a vacant room, label the image and upload the original alongside it, which AB 723 requires in California.
3. Description and ads: Draft from a fact sheet you wrote, then check each claim against the property and read the copy for language that signals a preferred buyer.
4. Lead response: A CRM or nurture tool can reply to inquiries within seconds, provided the lead consented and the tool honors "stop."
5. Showings and offers: Use AI for scheduling and for drafting offer summaries, and confirm every number against the documents.
6. Under contract: AI can summarize documents and draft deadline reminders, though a summary does not replace reading the contract.
7. After closing: Draft thank-you notes and review requests, and send them only to people who agreed to hear from you.

Writing Listing Copy With AI: A Fact Sheet and a Language Check

The most common AI task is the listing description, and it is also where fair housing and accuracy problems appear. Start with a fact sheet that you write, and give the assistant only those facts.

Write a 120-word listing description using only the facts below. Do not add features, measurements, school names, neighborhood claims or comparisons that are not in the facts.

Describe the property and its features, not the kind of person who should buy it. If anything I list is unclear, ask me a question first.

A useful fact sheet is short and documented:

• Basics: Address, bedrooms, bathrooms and the source of the square footage.
• Age and updates: Year built, plus the dates of roof, HVAC and other updates you can document.
• Features: Only what you have seen or can prove, such as a fenced yard or a finished basement.
• Costs and rules: HOA dues, taxes and any restrictions that apply.

Then read the draft against this check before it goes anywhere:

1. Who it suits: Remove phrases about who the home is "perfect for."
2. Schools, places of worship and demographics: Describe distances and amenities in neutral terms, or leave them out.
3. Unverified claims: Delete any feature, renovation or upgrade you cannot document.
4. Superlatives: Replace words such as "best" with facts a buyer can check.
5. Numbers: Confirm square footage, lot size and room counts against the MLS or county records.

An assistant can follow a prompt like this one, and it can still slip. The review step is where you catch it.

Also Read: AI Tools That Are Actually Changing How Professionals Work

The Tools, Grouped by the Job They Do

Reviewers rarely recommend one platform for everything. A common pattern is a CRM with built-in AI, a lead-nurture layer, a staging tool and a general assistant for writing.

For the CRM, reviewers most often name Follow Up Boss, Lofty, formerly Chime, and BoldTrail from Inside Real Estate, formerly kvCORE. They describe AI features such as lead scoring, follow-up drafts and conversation summaries.

Follow Up Boss is often described as a CRM that stays out of the way, while Lofty bundles a CRM with a website and lead tools.

Ylopo, whose assistant is called Raiya, and Structurely add an AI layer that replies to online leads by text and email, qualifies them and hands the conversation to the agent, according to reviews. Ylopo plugs into several CRMs.

These tools are the ones most affected by the consent rule above, so check how each records consent and handles opt-outs.

For photos, reviewers name Virtual Staging AI and REimagineHome, which work from a photo of an empty room, and they also mention Restb.ai. Whichever you use, plan to label the result and keep the original, as the photo rule requires.

For listing copy, email drafts and scripts, agents commonly use a general assistant such as ChatGPT, Claude or Gemini. The assistant writes from whatever facts you supply, which is why the fact-sheet habit above matters.

Each category has its own failure mode. CRM summaries can misstate what a lead said, nurture bots can quote a price or feature that is wrong, and staging tools can change how large a room looks. Read what a tool produces before a client does.

Pricing and bundles change frequently, so use the vendors' own pages for current figures, including those for Follow Up Boss, Lofty and Virtual Staging AI.

Also Read: The Best CRM Software for Startups: Matched to Your Stage and Budget

Seven Questions to Ask Any Vendor Before You Buy

A demo shows what a tool can do. These questions show what it does with your photos, your leads and your liability.

1. Original photos: Does the tool keep unaltered originals and help you label edited images?
2. Consent records: Can it store how and when each lead agreed to be contacted?
3. Opt-outs: Does it stop messaging when someone replies "stop" or a similar word?
4. Client data: Where is client information stored, and is it used to train models?
5. Fair housing review: Can you read and edit copy before anything is published or sent?
6. MLS and local rules: Does it support the staged-photo and advertising rules in your MLS and state?
7. Exit: Can you export your contacts, conversations and images if you cancel?

Keep the answers in writing. The duty to advertise truthfully rests with the licensee, whichever tool drafted the copy.

Also Read: How AI Chatbots Are Revolutionizing Customer Service

Where to Find the Rules for Your Market

Rules differ by state, MLS and brokerage, and they are changing quickly. These are the places to look, in the order most agents reach them.

• Your MLS: Photo, staged-image and advertising rules appear in its rulebook and in broker notices, as CRMLS and San Diego MLS published for AB 723.
• Your state real estate commission: Licensing law and advertising rules live here, and AB 723 sits inside California's licensing law.
• NAR and your local association: The Code of Ethics, including Articles 2 and 12, and NAR's AI guidance for brokers.
• Your brokerage: Ask for the current AI policy, since many brokerages set internal rules on which tools you may use.
• An attorney: Bring in a real estate attorney for anything involving calls, texts or tenant screening.

Keep a Short Paper Trail for Every Listing

If a photo, a message or a description is ever questioned, a small file answers most of the questions. Save these items for each listing:

• Original photos: Keep the unedited images next to every altered version.
• Labels: Save a screenshot of how each altered image was disclosed.
• Fact sheet: Keep the facts you gave the AI tool and the draft it produced.
• Consent records: Store how and when each lead agreed to be contacted.
• Vendor terms: Save the terms and settings in effect when you used the tool.

Store them with the listing file your brokerage already keeps, and ask your broker how long to retain them.

Conclusion

AI saves an agent the most time on drafting and follow-up, and it adds risk when its output is published or sent unchecked. The three rules cover most of that risk: fair housing language, disclosure of altered photos and consent for automated contact.

Sort your tasks against those rules, keep originals and consent records, and treat every AI draft as a draft. Agents who build those habits can use the tools freely, which is where the time savings are.

Also Read: The Best AI Image Generators for Creating High-Quality Visuals