You click the link. You see the headline. “Spotless.” “Popular.” “Rave reviews.”
It sounds like a vacation paradise.
Then you scroll.
The guest reviews tell a completely different story. One user found dead mice near their seat. Another got food poisoning from raw chicken. A wedding party of 68 people got sick.
This isn’t a glitch. It’s the new normal on Tripadvisor, and it’s dangerous.
A recent investigation by UK consumer watchdog Which? reveals that Tripadvisor’s new AI summary tools are systematically downplaying or ignoring severe health and safety risks. Instead of warning you away, these algorithmic summaries are painting a rosy picture that directly contradicts the harrowing experiences of real guests.
The Riu Palace Disaster
Take the Riu Palace Santa Maria in Cabo Verde. It’s a five-star, all-inclusive resort. Tripadvisor’s AI calls it “popular with many travelers” with “spacious rooms” and cleanliness described as “spotless.”
The reality? Guests report “exceptionally poor hygiene” and “no basic cleaning standards.”
Food there isn’t just bad—it’s dangerous.
One guest served raw chicken. Others photographed flies and birds crawling over buffet food. One woman found roasted mice by her sitting area. A family fell ill so severely they had to fly home early.
“This place will destroy holidays, and has the potential to take lives.” — Tripadvisor guest review
When Which? checked in March, there were 102 mentions of food poisoning in recent reviews for this specific hotel. Between December 2025 and April “2026” (likely a typo for 2024 or 2025 in source, but reported as is), 32 low-rated reviews noted serious illness. Seven deaths have been reported since 2023 at this property.
A group legal action involving 412 holidaymakers is underway.
And the AI? It still calls the place “spotless.”
Even the interactive trip planner, Ollie, fails. When asked directly if food poisoning is a risk at Riu Palace, the bot replied that it was “quite unlikely” and praised the resort’s “high hygiene standards.”
Minimizing the Worst Offenses
This isn’t isolated to one hotel in Cabo Verde. Tripadvisor’s AI consistently softens the language around serious safety breaches.
At the Garza Blanca resort in Cancún, recent reviews cited illnesses, including among a wedding group. The AI summary? “Immaculate cleanliness” and positive dining feedback.
The Occidental Caribe in the Dominican Republic received a similar treatment. Reviewers in March called it a “disaster,” noting sewage smells, mold, and zero running water. One guest showered with bottled water. The AI summary highlighted “abundant amenities” and only vaguely mentioned “inconsistent” cleanliness.
But the most disturbing cases involve human safety, not just bad food.
In Antalya, Turkey, at the Kaia Coracesium hotel, guests reported sexual harassment. Male staff made inappropriate jokes, followed guests to their rooms, and pressured women for social media details. One waiter followed a guest up the stairs.
Tripadvisor’s AI summarized the service as “friendly.” The only nod to these allegations?
“Lapses [in service] noted by few.”
Think about that phrasing. “Lapses.” “Few.” It turns sexual harassment and stalking into minor service errors.
How the AI “Thinks” (And Why It’s Broken)
Tripadvisor claims its AI uses large language models to identify common themes. It says it aims for transparency, showing both positive and negative feedback. Summaries update monthly based on the last 12 months of reviews.
But the execution is flawed.
The algorithm seems to prioritize volume over severity. If “clean” is mentioned frequently, even by a minority, it surfaces. If a single reviewer writes a graphic account of illness, the AI might miss it—or sanitize it—especially if other guests mention “good service” or “nice view.”
The language is also oddly generic. It avoids specific keywords like “food poisoning,” “sexual assault,” or “mold.” It prefers vague, neutral terms that don’t scare the user.
Is that the point?
To keep you clicking? To keep the ad revenue flowing? Or just because the AI lacks the nuance to understand context?
Google Does It Better
Here’s the kicker. Google’s AI summaries handle this significantly better.
For the same Riu Palace, Google’s overview warns of “potential for illness” and flags “outbreaks.” It mentions concerns about birds in buffet areas. It doesn’t hide the danger.
Compare this to the Britannia International in London, ranked worst by Which? for over a decade. Tripadvisor’s AI says guests “often praise clean rooms” and the atmosphere is “charming.”
Google says the hotel is “frequently rated as one of the worse hotel chains in the UK,” highlighting “filthy” conditions and “horrendous” service.
One AI summarizes the truth. The other summarizes the marketing.
The Danger of the “Easy Digest”
Tripadvisor’s spokesperson insists these summaries are not meant to replace individual reviews. They’re just a snapshot. Users should click through to see the full context.
Rory Boland, editor of Which? Travel, calls that advice inadequate.
“This ignores the fact that [Tripadvisor made] the decision to push these summaries to the top of the page,” Boland says. “This failure to surface critical safety information is unacceptable. And potentially life-threatening.”
He’s right. Most travelers won’t click through. Most will trust the summary. Most will book the “spotless” room.
Which? advises scrolling past the AI. Look at one-star reviews. Check other sites. Verify the food. Verify the safety. Verify everything.
Because the AI isn’t watching your back. It’s watching the sentiment.
And it’s missing the monsters in the buffet.
The reviews are there. The warnings are there. But they’re buried under layers of algorithmic gloss.
Will you dig them out?
