Key facts
- Tripadvisor's AI summaries for hotels have been found to downplay or omit serious safety issues, according to a UK consumer watchdog.
- The Riu Palace Santa Maria in Cape Verde was described as 'spotless' by Tripadvisor's AI, despite numerous guest reports of severe food poisoning, poor hygiene, and even a death.
- Tripadvisor's AI trip planning bot, Ollie, also incorrectly stated food poisoning was 'unlikely' at the Riu Palace.
- AI summaries for other hotels like Garza Blanca in Cancun and Occidental Caribe in the Dominican Republic also failed to reflect serious guest complaints about illness and poor conditions.
- Allegations of sexual harassment at Kaia Coracesium in Turkey were summarized by Tripadvisor's AI as mere 'lapses in service'.
- Google's AI summaries for the same hotels provided more accurate warnings about potential health risks and poor conditions.
Tripadvisor's AI-generated summaries at the top of hotel listings have been found to significantly downplay or omit serious safety concerns, including food poisoning, poor hygiene, and sexual harassment, according to an investigation by UK consumer champion Which?.
The five-star Riu Palace Santa Maria in Cape Verde was described by Tripadvisor's AI as 'spotless' with 'rave reviews' for its restaurants. However, recent guest reviews detail 'exceptionally poor hygiene,' 'awful, bland, unsafe and inedible' food, and numerous instances of food poisoning, with some guests hospitalized and one death reported this year. A group legal action is underway representing 412 holidaymakers who claim to have fallen ill at the property.
Tripadvisor's AI trip planning bot, Ollie, also incorrectly stated that food poisoning was 'quite unlikely' at the Riu Palace and that it had a 'strong reputation for high hygiene standards.' Similar issues were identified with AI summaries for Garza Blanca resort in Cancun and the Occidental Caribe in the Dominican Republic, which glossed over guest reports of illness, sewage smells, mould, and lack of running water.
Furthermore, at Kaia Coracesium in Turkey, AI summaries described 'friendly' service and 'lapses in service' for a hotel where multiple reviewers reported feeling unsafe due to repeated sexual harassment from staff. Tripadvisor stated its AI summaries use large language models to identify common themes from reviews and aim for transparency, but acknowledged it is investigating the examples provided by Which?.
In contrast, Google's AI summaries for the same hotels provided more accurate warnings, flagging 'potential for illness' and 'concerns over birds in the buffet areas' for the Riu Palace, and 'filthy conditions' for a Britannia hotel.
Tripadvisor maintains its AI summaries are not intended to replace individual reviews and that users can easily access the full traveler quotes. However, Which? argues that pushing these summaries to the top of the page without surfacing critical safety information is unacceptable and potentially life-threatening, advising users to bypass them and consult individual reviews.