A good photograph has always helped to sell a house. Estate agents have long relied on wide-angle lenses, careful lighting and the occasional fresh flowers to show a property at its best. Artificial intelligence has, however, taken this to a new level. In seconds, AI tools, such as ChatGPT, Claude and Gemini, can furnish an empty lounge, green a patchy lawn, swap a grey sky for a golden sunset or erase a crack in the wall. Buyers abroad increasingly report arriving at viewings only to find that the lush lawn in the listing is, in reality, mud. This raises a critical question: where does enhancement end and misrepresentation begin?
The US state of California was among the first to answer this question, with Assembly Bill 723 coming into effect on 1 January 2026. It requires estate agents who use digitally altered images to disclose the alteration on or next to the image, and to give buyers access to the original photograph through a website, URL or QR code. The law covers changes to furniture, flooring, paint colours, landscaping, window views and even neighbouring properties. Routine adjustments such as lighting, cropping and white balance are excluded. A breach may result in a misdemeanour charge and a fine of up to $2,500.
Australia has followed suit. On 24 September 2026, the New South Wales Parliament passed legislation requiring landlords and agents to disclose rental images that have been digitally altered or AI-generated in a way likely to mislead, such as fake furniture that hides damage or misrepresents room sizes. The penalties are up to AUD 5,500 for individuals and AUD 22,000 for businesses, and the law is expected to come into effect in early 2027. Notably, neither jurisdiction bans AI, but rather focus on disclosure where an edit changes how the property itself is represented.
The South African property market has been no stranger to fierce debates and rapidly changing legislation over the past few years. However, despite the speedy introduction of AI in our everyday lives, South Africa currently has no legislation, PPRA rule or guidance that specifically regulates the use of AI in property marketing. The Protection of Personal Information Act 4 of 2013 (“POPIA”) is often suggested as the answer. However, its transparency obligations apply to the processing of personal information, not to the accuracy of a photograph of a building. POPIA would only become relevant where listing photos show identifiable people or vehicle registration numbers, or where a client’s photos are uploaded to a third-party AI tool, which may process them offshore.
This does not mean that the law is silent. Section 41 of the Consumer Protection Act 68 of 2008 (“CPA”) prohibits a supplier from making a false or misleading representation about a material fact, or from failing to disclose a material fact where that failure amounts to deception. The section specifically covers representations that immovable property has characteristics, facilities or natural features that it does not have, and it extends to persons acting on the supplier’s behalf. Although the CPA applies mainly to sellers and landlords acting in the ordinary course of business, such as developers and portfolio landlords, the principle is clear.
Furthermore, the Code of Conduct under the Property Practitioners Act 22 of 2019 requires practitioners to exercise reasonable care and skill. It also contains a specific duty not to make misrepresentations or false statements, and a breach may constitute sanctionable conduct. The common law adds a further layer: a photograph is a representation. A buyer who purchased on the strength of an image that concealed a defect may claim rescission or damages for misrepresentation, and a voetstoots clause offers no protection to a seller who deliberately concealed a known defect. In one reported US dispute, a buyer relied on a digitally enhanced fireplace, only to discover a chimney defect costing up to $30,000 to repair. The altered photograph then became the buyer’s main evidence in a misrepresentation claim against the seller.
The practical line is therefore not whether AI was used, but whether the edit changes how the property is represented. Brightening a dark photo is unlikely to raise concern. Adding furniture to an empty room should be clearly labelled as virtual staging. Removing damp, cracks, power lines or a neighbouring building, however, crosses the line. Agencies would be well advised to label altered images, keep the originals on file, and ensure that every AI-generated image is checked against the property before it is published
Until South Africa adopts its own rules, the safest approach for agents is simple: if the photo shows something the buyer will not find, say so. Buyers and tenants, in turn, should treat listing photos as an introduction rather than a guarantee. Always view the property in person before signing, ask whether the images have been edited, and record any material feature in the offer or lease. After all, a picture may be worth a thousand words, but in a property transaction, it should not be worth a legal dispute.
Should you need assistance with the use of AI in your marketing, please contact our IT & IP department: 021 914 4020 or dataprotection@dkvg.co.za
| Cookie | Duration | Description |
|---|---|---|
| cookielawinfo-checkbox-analytics | 11 months | This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Analytics". |
| cookielawinfo-checkbox-functional | 11 months | The cookie is set by GDPR cookie consent to record the user consent for the cookies in the category "Functional". |
| cookielawinfo-checkbox-necessary | 11 months | This cookie is set by GDPR Cookie Consent plugin. The cookies is used to store the user consent for the cookies in the category "Necessary". |
| cookielawinfo-checkbox-others | 11 months | This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Other. |
| cookielawinfo-checkbox-performance | 11 months | This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Performance". |
| viewed_cookie_policy | 11 months | The cookie is set by the GDPR Cookie Consent plugin and is used to store whether or not user has consented to the use of cookies. It does not store any personal data. |