Tripadvisor AI smooths warning signs in hotel reviews

What this is about

Which? published an investigation on 2 July 2026 into Tripadvisor. The concern: AI summaries at the top of hotel reviews can smooth over serious complaints and give travellers an overly positive impression. The Guardian reported examples involving alleged food poisoning, hygiene problems, lack of running water and sexual harassment.

This is more than a travel annoyance. Platforms use AI summaries so people do not have to read hundreds of individual reviews. That makes weighting critical: a summary is not a neutral shortening, but a decision about what is important enough to surface.

What Tripadvisor Review Summaries actually do

The feature condenses many user reviews into short overviews. Tripadvisor says users can still click through and inspect the underlying reviews. In practice, though, the summary sits where many people look first.

Which? describes, among other examples, a Cape Verde hotel whose AI overview presented cleanliness and restaurants positively, while individual reviews reported raw food, insects near buffet food and illness. In other cases, warnings about water problems or harassment were allegedly weakly surfaced or not visible enough.

Why it matters

For hotels, bad information is not only inconvenient. It can affect money, health and safety. If an AI turns many reviews into a friendly average, outliers lose force. Yet those outliers are often decisive when the topic is hygiene, violence, fraud or safety risk.

Tripadvisor disputes Which?'s framing and says review content is not suppressed. The company points to safeguards and stresses that AI summaries do not replace individual reviews. Still, the case exposes a structural problem: users quickly treat prominent summaries as a verdict.

In plain language

Imagine asking five friends about a restaurant. Four say, "It was fine," and one says, "I got sick afterwards." A bad summary would be: "Most people found it fine, with minor issues mentioned." That sounds balanced. In practice, it hides the warning you most needed.

AI summaries work in a similar way. They can describe the average well and still lose the information that would have changed your decision.

A practical example

A family plans a ten-day holiday costing EUR 3,200. The hotel has 2,000 reviews and an AI summary that praises rooms, location and food. Among the latest 80 reviews, however, 12 mention stomach illness and several point to poor hygiene.

If the family reads only the AI box, the risk looks small. If they check one-star reviews and recent photos, they see a different picture. The practical difference is not academic: it affects booking, insurance, cancellation and health.

Scope and limits

• The investigation does not prove that every Tripadvisor summary is wrong; it shows problematic examples.
• Individual reviews can themselves be mistaken, exaggerated or manipulated; they also need context.
• AI summaries are especially risky when platforms introduce them as a convenience feature without clearly showing which warning signals were prioritised.

SEO & GEO keywords

Tripadvisor, Which?, AI summaries, hotel reviews, travel platforms, consumer protection, online reviews, travel safety, platform trust, review summarisation