AI Compliance & Disclosure: What Real Estate Agents Need to Know

The Regulatory Moment

AI adoption in real estate has outpaced regulation by about two years. That gap is closing fast.

California's AB 723 took effect January 1, 2026. It requires real estate licensees to disclose when listing photos or videos have been materially altered using AI. Virtual staging, sky replacements, landscaping enhancements, decluttered rooms—if AI changed what a buyer sees, you must say so.

This is the first state law specifically targeting AI use in real estate listings. It will not be the last.

At the same time, the National Association of Realtors updated its guidance on AI use in late 2025, HUD reaffirmed that the Fair Housing Act applies to AI-generated content, and multiple MLS organizations have added AI disclosure fields to their listing input forms.

The message is clear: use AI, but use it transparently.

California's AB 723: What You Need to Know

AB 723 is the most significant AI regulation to hit real estate so far. Here are the specifics.

What It Requires

• Disclosure of material alterations: Any listing photo or video that has been "materially altered" using AI or digital editing tools must include a clear disclosure.
• "Material" means perception-changing: If the alteration could affect a buyer's understanding of the property's actual condition, it requires disclosure. Virtual staging, sky replacements, removing power lines, enhancing landscaping, fixing visible damage in photos—all qualify.
• Clear and conspicuous placement: The disclosure must appear adjacent to the altered media. A buried footnote on page 12 of the listing packet does not count.
• Applies to all marketing channels: MLS listings, social media posts, property websites, email campaigns, print materials. If it contains AI-altered imagery and markets a property, it needs a disclosure.

What It Does Not Require

• Disclosure of basic photo editing (brightness, contrast, color correction)—standard photography adjustments are exempt.
• Disclosure for AI-generated text (listing descriptions, emails)—the law specifically targets visual media.
• Use of specific disclosure language—the law sets the standard ("clear and conspicuous") but does not mandate exact wording.

Penalties

Violations fall under the California Department of Real Estate's disciplinary authority. Consequences include:

• Formal citation and fine
• License suspension or revocation for repeat violations
• Civil liability if a buyer claims they were misled by undisclosed AI alterations

The fine for a first offense is not the real risk. The real risk is a buyer who discovers the "beautifully landscaped yard" was AI-generated, feels deceived, and files a complaint with DRE while simultaneously pursuing a civil claim. That combination can end a career.

State-by-State Regulatory Landscape

California moved first, but it is not alone. Here is where other states stand as of early 2026.

The pattern: States with large real estate markets and active consumer protection frameworks are moving first. Even in states without specific legislation, existing consumer protection and deceptive practices laws can apply to misleading AI-generated content. Do not assume you are exempt because your state has not passed an AI-specific bill.

Fair Housing Meets AI

AI compliance is not just about disclosure. It intersects directly with Fair Housing law—and this is where the stakes get highest.

AI-Generated Text

We covered this in depth in our Fair Housing compliance guide, but the core issue bears repeating: AI will confidently generate discriminatory language. "Perfect for young families." "Safe, quiet neighborhood." "Close to the community church." Every one of these is a Fair Housing problem, and AI produces them by default.

HUD's guidance is explicit: housing providers are responsible for AI-generated content. "The algorithm did it" is not a defense. First-offense fines start at $21,663. Jury awards have exceeded $2 million.

AI Virtual Staging Bias

This is the emerging risk most agents are not thinking about. When AI virtually stages a home, it makes choices about furnishings, art, decor, and lifestyle cues. Those choices can signal demographic preferences.

Examples of AI bias in virtual staging:

• Staging every home with decor that signals a specific cultural or religious background
• Including family photos or lifestyle imagery that depicts only one demographic
• Using furnishing styles that consistently signal a particular income level or age group
• Placing religious items, cultural artifacts, or demographically-coded elements

Is a single staged photo going to trigger a Fair Housing complaint? Probably not. But a pattern of AI-staged homes that consistently depict one demographic creates the kind of evidence that Fair Housing attorneys look for.

AI Ad Targeting

Meta (Facebook) paid $115,000 to settle allegations that its ad platform allowed housing advertisers to exclude users by race, religion, and other protected classes. AI-powered ad targeting can create the same problem even without explicit exclusion, by optimizing for audiences that correlate with protected characteristics.

If your AI ad platform "optimizes" your listing ads toward a narrow demographic, you may be engaged in discriminatory advertising without knowing it.

Best Practices: Audit Trails and Disclosure Templates

Build an Audit Trail

Documentation is your best defense. For every piece of AI-generated or AI-altered content, record:

This takes 60 seconds per content piece. It creates a documented record of responsible AI use that demonstrates good faith if a question ever arises.

Disclosure Templates

Do not reinvent the wheel for every listing. Build templates and use them consistently.

Adapt these to your state's requirements and brokerage policies. The key elements: state that AI was used, specify what was altered, note that it may not reflect current conditions, and point toward verification.

The OODA Loop as a Compliance Framework

We teach the OODA Loop as a verification framework for all AI output. It works particularly well for compliance because it is systematic and repeatable.

Observe: Inventory Your AI Touchpoints

Before you can comply, you need to know where AI shows up in your workflow. Walk through a typical listing and catalog every AI touchpoint:

• Listing description—written or assisted by AI?
• Photos—virtually staged, enhanced, or sky-replaced?
• Market analysis—AI-generated comparables or narrative?
• Social media posts—AI-drafted captions or graphics?
• Email campaigns—AI-written subject lines or body copy?
• Ad targeting—AI-optimized audience selection?

Most agents discover they use AI in more places than they realized. That is fine. You just need to know where.

Orient: Map Regulations to Touchpoints

For each AI touchpoint, identify which regulations apply:

• AI-altered photos/videos: State disclosure laws (AB 723 in CA), MLS rules
• Listing descriptions: Fair Housing Act, state deceptive practices laws
• Market analyses: Accuracy obligations, fiduciary duty
• Ad targeting: Fair Housing advertising rules, platform policies
• Client communications: Agency disclosure requirements, fiduciary duty

Decide: Classify and Prioritize

Not all AI use carries equal risk. Prioritize by potential consequence:

• High risk (address immediately): AI-altered listing photos without disclosure, AI-generated descriptions with Fair Housing violations, AI ad targeting that excludes protected classes
• Medium risk (address in workflow): AI market analyses without verification, AI social content without review, AI emails with unverified claims
• Lower risk (build into process): AI-drafted internal documents, AI scheduling and organization, AI research summaries

Act: Execute and Document

For each high-risk item, take action now. Add disclosures to existing listings. Review descriptions for Fair Housing language. Check ad targeting settings. Log everything in your audit trail.

Then cycle back to Observe. Compliance is not a one-time event. It is a continuous loop—which is exactly why the OODA framework fits.

NAR Guidelines on AI Use

The National Association of Realtors has been developing AI guidance since 2024. Here is where their recommendations stand:

• Transparency: Disclose AI use to clients when AI-generated content is presented as part of your professional services. This includes AI-written CMAs, listing descriptions, and marketing materials.
• Accuracy: Verify all AI-generated data, statistics, and property information before publishing or sharing with clients. AI hallucinations are your liability.
• Fair Housing: Review all AI-generated content for Fair Housing compliance before publication. Implement AI guardrails in your prompts.
• Client data: Do not input confidential client information into public AI tools. Use enterprise or privacy-compliant versions when handling sensitive data.
• Professional judgment: AI assists your expertise—it does not replace it. Maintain your professional judgment and fiduciary responsibility regardless of what AI suggests.

NAR's position is practical: AI is a tool, not an excuse to abdicate professional responsibility. Use it. Disclose it. Verify it.

MLS-Level Requirements

Many MLS organizations are implementing their own AI policies, often ahead of state law. Common requirements appearing across MLSs:

• AI photo flags: Checkbox or dropdown field to indicate virtual staging or AI enhancement
• Disclosure text fields: Dedicated space for AI disclosure language in listing remarks
• Original photo requirements: Some MLSs now require at least one unaltered exterior photo alongside AI-enhanced images
• Description attribution: Emerging requirement to note AI-assisted descriptions

Check your local MLS rules. They may already have requirements you are not following.

Building a Compliance-First AI Workflow

Compliance should not be an afterthought bolted onto your AI workflow. Build it in from the start.

This process adds about five minutes to your listing workflow. The alternative is unlimited legal exposure.

Frequently Asked Questions

Do I have to disclose AI-generated listing photos?

In California, yes—as of January 1, 2026, AB 723 requires it. In other states, check your local MLS rules and state consumer protection laws. Even where not legally required, NAR recommends voluntary disclosure. The direction of regulation is clear: disclosure will become standard everywhere. Getting ahead of it now protects you later.

What is California's AI disclosure law for real estate?

AB 723 requires California real estate licensees to provide clear, conspicuous disclosure when listing photographs or videos have been materially altered using AI or digital tools. "Material" means changes that affect a buyer's perception of the property's condition—virtual staging, sky replacement, landscaping enhancement, damage removal. Basic photo adjustments like brightness and contrast are exempt. Violations can result in DRE disciplinary action including fines and license suspension.

Can AI-generated content violate fair housing laws?

Absolutely. AI text can include discriminatory language describing who should live in a property. AI virtual staging can depict furnishings and lifestyle cues that signal preference for certain demographics. AI ad targeting can create discriminatory audience segments even without explicit exclusion settings. HUD has confirmed that housing providers bear full responsibility for AI-generated content. Intent does not matter. If the output is discriminatory and you published it, you are liable.

What should my AI disclosure template look like?

Your disclosure should state that AI was used, specify what was altered (virtual staging, sky replacement, etc.), note that images may not represent the property's current condition, and direct buyers to verify by visiting the property. Place the disclosure directly adjacent to altered media—not buried in fine print. See the templates earlier in this article for ready-to-use language.

Do I need an audit trail for AI-generated real estate content?

It is not legally mandated in most states yet, but it is strongly recommended and may become required. An audit trail documents which AI tools you used, what content was generated, what edits were made, and who reviewed it for compliance. This record is your evidence of due diligence. If a complaint is filed, the difference between "I have documentation of my review process" and "I just used ChatGPT and posted it" is the difference between a warning and a serious disciplinary action.