← All guides Property

Are these Airbnb photos real?

Virtual staging, replaced skies and rooms that were never that wide. What a pixel check finds in a holiday rental listing, and the four questions it cannot answer.

· 9 min read · Best AI Image Detector

Most rental listing photos are real rooms that have been enhanced rather than invented. The useful finding is localised: a replaced sky, a removed object, or furniture added to an empty room, all of which show on the region map rather than in the headline score.

What listing photography actually does

Short-let photography is a small industry with its own conventions, and almost all of them push a score upward without any dishonesty being involved. Wide lenses, bracketed exposures merged into one frame, lifted shadows and boosted colour are standard.

On top of that sits a layer of editing that is now routine: a grey sky replaced with a blue one, a bin removed from the kerb, a stain lifted from a carpet, a reflection of the photographer erased from a mirror.

Then there is virtual staging, where an empty or sparsely furnished room is filled with generated furniture. This is disclosed on some platforms and not on others, and it is the single largest source of the gap between a listing and an arrival.

None of these is the same as inventing a property, and conflating them produces the wrong response. A guest who arrives to find the sofa is a different colour has a complaint; one who arrives to find the building does not exist has a crime.

What the region map surfaces

Every one of those edits is localised, which is exactly the shape the region map is built to read. The whole-frame score reflects mostly-real pixels and stays unremarkable; the grid shows where the software worked.

16 14 18 15 21 19 17 20 74 81 77 22

The walls, windows and ceiling are photographed. The bottom row, where the furniture sits, is not.

A photograph of a real room with generated furniture placed in the lower half. Whole-frame score: 38.

Read the position of the heat rather than its magnitude. Heat along the top of the frame is usually a replaced sky. Heat in one corner is usually a removed object. Heat across a whole floor plane is usually staged furniture.

A uniformly warm grid with no structure to it is the least interesting result, because it is the signature of heavy global editing rather than of any specific change. That describes most professional listing photography.

The four questions a check cannot answer

This is where most of the risk in a rental booking actually lives, and none of it is visible in pixels.

  • Is this the right building? A real photograph of a real apartment somewhere else scores as the genuine photograph it is.
  • Is the host real? Listing fraud usually reuses photographs from a legitimate property, often in another city.
  • Is it as big as it looks? Wide lenses distort volume, and no detector measures a room.
  • Is it still like this? A genuine photograph from four years ago is genuine and out of date.

The first two are reverse image search problems. Running the main listing photograph through an image search and finding it on an estate agent's page in a different country is the single most valuable check available, and it takes ten seconds.

The third and fourth are questions for the host. Ask for a short video walkthrough taken today, or a photograph from a specific corner you nominate. A real host with a real property can do both; a fabricated listing cannot.

What to do with each kind of finding
What you seeLikely explanationReasonable response
Heat along the top edgeSky replacedCosmetic, ignore
Heat on one small patchObject removedAsk what was there
Heat across the floorVirtual stagingAsk if furniture is included
Warm everywhere, flatHeavy professional editingNormal for the category
High score, flat gridWholly generated roomDo not book

A proportionate booking routine

Scale the effort to the money and to how reversible the booking is. A refundable two-night stay booked through a platform with a guarantee needs almost none of this. A month-long let paid directly to a host needs all of it.

  1. Reverse image search the hero photo

    A match on an estate agent listing or a hotel site elsewhere ends the enquiry immediately.

  2. Check two or three interior shots

    Read the maps for staging and removed objects, and see whether the rooms are consistent with each other.

  3. Look for a photograph nobody would fake

    The corridor, the stairwell, the view from the actual window. Fabricated listings are all hero shots.

  4. Ask for something specific

    A video walkthrough, or a photo from a corner you nominate. This defeats every version of the problem at once.

  5. Pay through the platform

    A request to move payment off-platform is a stronger signal than anything in any photograph.

Reading a set rather than a picture

One listing photograph is a weak test. A set of them is a much stronger one, because a fabricated listing has to keep an entire property consistent across a dozen images and a real one does that for free.

Run three or four and look at the spread. Photographs taken in one session by one photographer land in a similar range, because they share a camera, a lens and an editing pass. A set assembled from different sources scatters, and that scatter is the finding.

Look at the rooms for agreement too. Does the light come from the same side of the building in every shot, do the floors continue plausibly from room to room, does the view from one window match the view from another. Fabricated sets fail these long before any individual image does.

The absent photograph is also informative. Real hosts photograph the awkward parts eventually: the small bathroom, the stairs, the kitchen corner. A listing composed entirely of wide, flattering hero shots is either professionally staged or assembled from somewhere else.

What to do when you arrive

If the property does not match the listing, photograph the discrepancy before you unpack. Photographs taken on arrival, with their own timestamps, are the evidence a platform dispute actually turns on, and they are far more persuasive than an argument about the original images.

Report through the platform the same day rather than after the stay. Most guarantee schemes have a short window measured in hours, and a complaint raised on departure is usually too late regardless of how good the evidence is.

Questions people ask

Are AI-generated rental listings common?
Wholly generated properties are still uncommon. What is common is real photographs of real rooms with substantial editing, and photographs taken from a genuine listing elsewhere. The second is the one that costs people money, and it is a reverse image search problem rather than a detection one.
What is virtual staging?
Furniture and decor added to an empty or sparse room in software. It is widespread in property marketing, disclosed on some platforms and not others, and shows on a region map as heat across the floor plane while walls and windows stay cool.
Why does an ordinary listing photo score in the forties?
Because listing photography is heavily processed by convention. Bracketed exposures merged into one frame, lifted shadows, boosted colour and wide-angle correction all rewrite pixels, and that is the signal the model reads. A flat, warm grid is normal for this category.
How do I tell a replaced sky from a generated room?
By where the heat sits. A replaced sky produces heat along the top of the grid with the rest cool. A generated room produces heat across the whole grid evenly, because every part of it came from the same process.
The photos look real but the place was nothing like them. What happened?
Usually a wide lens and an old photograph rather than manipulation. Neither is detectable in pixels: a genuine picture taken four years ago on a very wide lens is genuine. Ask for a dated video walkthrough before committing to a long stay.
Is it worth checking every listing?
No. For a short refundable booking through a platform with a guarantee, the platform is doing the work. Save this for direct bookings, long stays, large deposits and any listing where the host wants payment outside the platform.