One in five hotel photos on major booking platforms shows indicators of being AI-generated or AI-enhanced. That is the finding of an analysis of 25,550 images used by hotels to advertise their rooms, facilities, restaurants and amenities. The study was conducted by Berlin-based marketing agency ABCD Agency in partnership with German forensic AI verification provider ContentGuard.me.

Hotel photos are one of the most decisive factors when booking accommodation online. Yet there are growing signs that the imagery travelers rely on is no longer just traditionally retouched but increasingly created or altered using artificial intelligence.

In a sample of 100 randomly selected hotels each across seven destinations a total of 25,550 hotel photos were analysed in May 2026. Among those popular summer holiday destinations like Crete, Mallorca, Sicily and Alanya.

The result: approximately 19% of all images – a total of 4,778 photos – contained at least one signal typically associated with AI generation or AI editing.

These signals can appear in technical markers such as file metadata, or as visual anomalies within the image itself – including inconsistencies, irregular pixel patterns or detail errors.

Crete Tops the List of Suspected Cases of AI

Among classic holiday destinations, the sample revealed a significant spread. The Greek island of Crete had the highest rate of flagged photos: 23% of analysed images – 960 out of 4,139 – showed at least one indicator of possible AI involvement. That means nearly one in four hotel photos on Crete raised a red flag.

Mallorca came out comparatively well, with around 9% of photos flagged (360 out of 4,014). In the Turkish resort town of Alanya, the figure was 13% – roughly one in eight images. The Italian island of Sicily came in at 11%, or approximately one in nine.

Why This Matters: Photos Drive Decisions Worth Thousands

Booking a trip has never been easier. A few clicks can commit travelers to holidays costing several thousand euros, pounds or dollars. Many consumers compare options quickly, make decisions under time pressure and rely heavily on imagery.

When hotel photos are AI-optimised or AI-generated, expectations can be distorted in critical ways:
Room size & proportions: AI enhancement can amplify wide-angle effects, making rooms, pools and facilities appear larger than they actually are.
Furnishings & condition: Furniture, bathrooms and outdoor areas can appear newer, cleaner or more upscale than they are in reality.
Surroundings & atmosphere: Lusher vegetation, clearer skies or extreme blue water can create an idealised impression that rarely matches what guests find on arrival.

These enhanced images set expectations that reality often cannot meet. The result: disappointment on arrival, complaints and a long-term erosion of trust in both platforms and hotels that depend on credible visual content.

High Rates for City Destinations

The analysis also covered two city destinations in Germany and the results were striking. In Hamburg, 36% of analysed photos were flagged (1,191 out of 3,285). In Berlin, the figure was 27% (1,017 out of 3,735).

One possible explanation: city hotels frequently use supplementary images that don’t show the property itself, but instead feature iconic landmarks – such as Berlin’s TV Tower. These kinds of scenes may be more likely to be AI-generated and could have skewed the results upward.

However, the core concern remains: whenever photos of hotel rooms, bathrooms, lobbies or amenities are AI-enhanced, guests risk paying for an experience that doesn’t match reality.

Jens Kramosch, deepfake expert and founder of ContentGuard.me: “Hotel photos have always been retouched to some degree. But AI takes this to a completely different level. When images are no longer just polished but fundamentally altered, it risks crossing the line from marketing into deception. Travelers need to look more closely and platforms need to start labeling AI-edited content transparently.”

Robin Wilfert, Founder of ABCD Agency, warns: “When you book a trip, you’re buying a promise. The photos that travelers use to choose a hotel are the most important part of that promise. AI must not be allowed to create a gap between what’s shown and what’s real.”

Methodology
In May 2026, 100 randomly selected hotels were analysed in each of the seven destinations listed above (700 hotels in total). A total of 25,550 hotel photos (no video content) were examined using AI-powered forensic analysis by ContentGuard.me. The analysis evaluated both technical indicators in file metadata and visual anomalies within the images themselves, such as inconsistencies and irregular pixel patterns. A photo was flagged as a suspected AI case when at least one signal or marker was detected. This analysis does not distinguish between fully AI-generated images and AI-enhanced images.