Holidaymakers are being urged not to rely solely on AI-generated hotel review summaries after a Which? investigation found that some of Tripadvisor’s automated overviews appeared to downplay serious health and safety concerns raised by guests.
The consumer watchdog examined Tripadvisor’s AI-generated review summaries, introduced in 2023 to provide travellers with a quick snapshot of guest feedback, and found several cases where overwhelmingly positive descriptions failed to reflect repeated reports of food poisoning, poor hygiene and other serious complaints.
Food poisoning reports buried beneath positive reviews
One of the most striking examples involved the Hotel Riu Palace Santa Maria in Cape Verde. While Tripadvisor’s AI summary described the five-star resort as having spacious rooms, diverse restaurants and excellent cleanliness, Which? uncovered 102 guest reviews mentioning food poisoning.
Guest reviews cited by Which? described raw chicken being served, poor hygiene standards and reports of flies and birds around buffet food. Several guests said they had cut their holidays short after becoming unwell, while others claimed entire families were affected.
The resort is currently the subject of legal action involving more than 400 holidaymakers who allege they became ill after staying there.
The allegations have not been proven in court. Riu Hotels & Resorts has denied the claims, stating that its Cape Verde hotels operate under strict international health and hygiene standards, with regular external inspections and food safety checks.
Which? also identified similar concerns elsewhere. At a resort in Cancun, Mexico, Tripadvisor’s AI highlighted “immaculate cleanliness” despite multiple guests reporting illness.
Another hotel in the Dominican Republic received glowing AI praise for its facilities even though reviewers described major sanitation issues and widespread sickness among guests. In Turkey, the AI summarised staff as “friendly” despite reviews alleging repeated sexual harassment by employees.
AI convenience shouldn’t replace common sense
The investigation also tested Tripadvisor’s AI travel assistant, Ollie. When asked directly about the Cape Verde resort, the chatbot reportedly told researchers that food poisoning was “quite unlikely” and that the hotel had a “strong reputation for high hygiene standards”.
Tripadvisor said its AI summaries are designed to help users digest large volumes of reviews by highlighting the broad balance of positive and negative feedback rather than replacing individual guest comments.
The company added that reviews remain fully visible on the platform and said it is actively refining its AI tools after being alerted to examples where summaries did not accurately reflect guest experiences.
Consumer experts say the findings serve as a timely reminder that artificial intelligence should be treated as a starting point rather than the final authority when planning a holiday. Reading recent one-star reviews, comparing feedback across multiple travel platforms and looking for recurring complaints can provide a much clearer picture of a hotel’s standards before booking.
