How Should Realtors Disclose AI-Edited Listing Images in 2026?

What an AI staging disclosure actually means

An AI staging disclosure is a clear statement that a listing image was digitally altered with artificial intelligence rather than photographed exactly as the property appeared on the marketing date. It should identify the material change, such as adding furniture, removing stored objects, changing room dimensions, replacing furnishings, opening curtains, cleaning clutter, or changing the exterior. Merely saying “professionally staged” may be inaccurate if no physical staging occurred, while “virtually staged” can be too vague if viewers are not told what changed. As of September 28, 2026, there is not one nationwide US rule requiring a single disclosure phrase for every AI-edited real-estate image. Requirements can come from state law, local ordinances, brokerage policy, listing agreements, MLS systems, advertising rules, and applicable representations about the property.

Also worth reading: Do Real Estate Agents Need to Disclose AI-Generated Listing Photos and Videos? · What Are the Virtual Staging Compliance Rules for AI-Generated Listing Images in 2026? · Is AI Virtual Staging Worth the Cost for Realtors in 2026?

The safest disclosure distinguishes conventional production editing from generative or structural alteration. Cropping a photo, correcting white balance, brightening a window, or removing a visible camera reflection may not materially change the property and may be ordinary photography practice. Adding a sofa, deleting a damaged floor, changing a kitchen cabinet, altering a window view, or making a small room look larger changes what the buyer sees and deserves specific attention. A useful disclosure therefore names the alteration, states that it is an AI-generated representation, and makes clear that the image is not a record of the room’s current physical contents. It should not imply that every image must be labeled identically regardless of the degree or purpose of editing.

Why real-estate photos need special care in 2026

Photos now function as property representations, not simply advertising illustrations. Buyers use them to decide which homes to tour, evaluate room size and condition, compare properties remotely, and sometimes begin legal disputes after closing. AI tools can make an unoccupied room look furnished in seconds, which is useful when the property itself is the selling point. Yet realistic-looking furniture can also give buyers a misleading expectation about scale, storage, sunlight, circulation, or what they will receive at closing. The ethical problem is not that AI is used; it is that an altered image can blur the line between presenting the property and inventing features.

The regulatory environment is moving. Public reporting in 2025 and 2026 described New York City proposals or requirements aimed at disclosing digitally manipulated listing images, prompting Realtor.com to ask whether other cities could follow. That does not mean every jurisdiction adopted the same rule by September 2026, and it does not establish that a proposed bill was enacted. Agents should check the current status rather than assume that the headline describes an active citywide mandate. Federal advertising rules can still matter when an image or generated description omits a material fact, while MLS rules may impose disclosure, accuracy, or alteration requirements even where no AI-specific statute exists.

A practical threshold is materiality: would a reasonable buyer rely on the changed detail when deciding whether to offer? If the answer is yes, disclose it. This test is more durable than asking whether a software vendor calls an operation “AI.” Generative fill, object insertion, image expansion, and virtual furniture are clearly alterations, but automated sharpening and color correction sit closer to routine retouching. Keep the original, edit history, and written record so the brokerage can demonstrate what was changed and when.

What to include in a clear AI staging disclosure

A strong disclosure should contain four elements: the words “AI” or “artificial intelligence,” identification of the altered image, a plain description of the change, and a statement that the result is illustrative rather than an actual photograph of the furnished property. For example, “AI virtually staged: the living-room image shows an AI-added sofa, rug, artwork, and plants; these items are not included and the room was empty when photographed” gives buyers more useful information than “AI image” alone. The wording should appear in the listing, near the relevant image or gallery, and wherever a cropped version is reused in social advertising.

Do not rely solely on a disclosure buried in a terms-of-use page, MLS disclaimer, or website footer. Buyers may view the image through a search-results thumbnail, a social post, an email campaign, or a saved brochure, where a general footer can be missed. Labeling an individual image as “AI virtually staged” is more direct. If only one room is altered, say which room; if six exterior shots have clouds removed, identify that processing. If a virtual scene adds architectural features, disclose those features rather than merely calling the picture “enhanced.”

The record should also state who approved the edit, the software category used, the date produced, and whether the image has been reviewed for physical accuracy. That does not require publishing every production detail, but it helps a broker respond to questions and separate ordinary color correction from substantial manipulation. Keep the unmodified source photograph for at least as long as local brokerage, transaction, and litigation-retention policies require. If there is no fixed period, a practical default is to preserve the original through the listing period, transaction, and any dispute-resolution process rather than deleting it after upload.

A practical workflow before publishing listing images

Begin with the property and not the desired result. Photograph the actual room, retain an unedited original, and note the date and time. Decide which furnishings are physically present before applying virtual changes. Agents should then create a written alteration record describing each generated or removed element and obtain brokerage approval before publication. This usually takes less than 15 minutes for a five-image property when a reusable form is used, although reviewing every image carefully can take 30 to 60 minutes for a large luxury or commercial listing.

Next, compare the edited image with the original at full size. Check whether the AI changed doors, windows, ceiling height, floor level, outlets, trim, railings, room width, or the apparent direction of natural light. Virtual staging can accidentally alter fixed features, and furniture may block or distort architectural details. A second reviewer should inspect the result because the person operating the software may focus on aesthetics rather than measurable accuracy. Reject the image if it makes the property appear to have an amenity, structure, storage area, or square footage that does not exist.

Publish the labeled image with its disclosure nearby, carry that label into derivative formats, and revise any captions that describe the virtual items as included. Retain the original, edited version, disclosure, and approval trail in the listing file. Revisit the language when tools or local rules change; a disclosure that is accurate in January may become incomplete if later versions substantially restructure a room. The workflow is not about certifying AI output as “perfect.” It is about controlling the process, identifying meaningful changes, and ensuring recipients can distinguish a real room from a proposed presentation.

AI virtual staging versus physical staging and conventional editing

Physical staging still has advantages. It can show scale accurately, let buyers understand natural light, and give the listing a coherent presentation without synthetic detail. It also costs more and consumes space; a vacant or occupied home may not be ready for a stager. Conventional editing is usually the least disruptive option because it corrects technical image problems while preserving the photographed scene. It becomes risky when “cleanup” becomes a material reconstruction, so the boundary depends on what changed and what a buyer might infer.

Cost figures vary by market, subscription, order minimum, and output count. A low-cost entry tool may provide a limited number of generations, while a managed service may charge per room, listing, or campaign. The 2026 range of roughly $10–$100 for exports or $20–$100 per month for entry-level tools is a planning estimate, not a promised price. Physical staging commonly runs from several hundred to several thousand dollars and rises sharply in major markets or for large homes. Compare the cost with the listing’s photography budget and the risk of losing buyer trust; the cheapest image is not necessarily the best business decision.

Common disclosure and image-editing mistakes

The most common mistake is vague labeling. “Representation,” “enhanced,” “render,” and “for illustration” can leave the buyer unsure whether furniture is virtual. Another error is disclosing only on the listing page while using unlabeled crops in social media, email, or paid advertising. Some agents remove visible evidence of alteration, such as a tiny “AI” icon, because it appears unattractive or may reduce engagement. That is poor practice: the disclosure is meant to prevent misunderstanding, not merely satisfy a platform badge.

A second mistake is allowing AI to change permanent features. Real-estate tools may regenerate windows, flooring, molding, appliances, cabinetry, and room geometry. If the edited image differs from the appraisal, measurements, or written property description, a generic staging label may not be enough. Compare AI images with floor plans and inspection information, and never use virtual furnishings to conceal material defects such as water damage, cracks, missing panels, or undesirable views without accurate disclosure. Cosmetic enhancement should not make a known condition visually disappear.

Do not promise “pixel-perfect accuracy” merely because the result looks realistic. Generative systems can still alter texture, reflections, shadows, object dimensions, and connections between spaces. Avoid using AI to add views of neighboring land, a park, water, or city scenery when the image merely simulates what might be seen from a different location. Finally, do not use another person’s listing, identity, logo, or copyrighted photograph without permission. AI disclosure does not cure copyright, privacy, fair-housing, or listing-accuracy problems.

When agents should act and which rules to check first

Act before the listing goes live, not after a buyer complains. The September 28, 2026 date makes an immediate review reasonable because the market, tools, and reporting around disclosure are changing. Start by asking the managing broker for the current written policy and checking the MLS’s image-attestation or virtual-staging fields. Then confirm whether the property is located in New York City or another jurisdiction with an AI-image rule, and verify the status of the law directly with an official legislative, regulatory, or brokerage source rather than relying on a news summary.

Agents should also check state advertising and brokerage rules, the listing agreement, and any applicable MLS attestation. If the listing uses multiple systems, apply the strictest relevant standard because compliance cannot be repaired merely by saying the primary website contains a disclaimer. When legal status is unclear, disclose the material AI change as a conservative business practice and seek written broker or counsel guidance. This is especially important for short-term rentals, new developments, occupied homes, luxury properties, vacant units, and any image depicting a view or amenity that affects value.

There is no universal need to abandon AI virtual staging. For an empty room that clearly receives virtual furniture, AI can save time, reduce physical disruption, and show a plausible use of space. The decision should pass four tests: the home remains recognizable, permanent features are unchanged, the material alteration is labeled, and the listing description remains truthful. If any test fails, return to the original image, use physical staging, use conventional editing, or hire a qualified photographer. A transparent image is usually more valuable than a technically impressive image that creates a dispute after the showing.

The final verification pass for compliant listing presentation

Before approval, view the gallery as a first-time buyer would. Every materially altered image should be identifiable without opening a footnote, and every caption should distinguish included furnishings from suggested ones. The disclosure should identify the property, the altered room or feature, the date the image was created, and the fact that the visual is AI-assisted. If the image was later exported or resized, confirm that the disclosure travels with it. The broker should sign off on the final set and archive the original and supporting records.

This process is manageable. For a typical five-image listing, a careful review and documented disclosure can often be completed in under an hour, while a larger 20-image property may require two to four hours. Those are workflow estimates, not legal deadlines. The important control is that the listing does not go live until a person has compared the final images with the originals, the written room dimensions, and the known condition of the property. Automation can help organize files, but it should not make the final compliance decision.

The definitive approach for 2026 is not a single magic phrase. It is an accurate, visible statement attached to the relevant image, backed by original files and a clear production record. Use “AI virtually staged” as a concise label only when the picture is retained and truthful, and add a sentence naming the generated items or changes. Recheck local rules at the time of publication because public discussion of a disclosure mandate does not prove that a proposal became law. Where the requirement is uncertain, the safer practice is to disclose rather than speculate, while avoiding claims that ordinary brightness or cropping always requires an AI-specific label.