Alex Rivlin thought he was pretty tech-savvy. The Las Vegas real estate developer had used AI for years and had come to trust it. But when he needed a shuttle number for his Royal Caribbean cruise last August, he did what millions of people do every day: he asked Google’s AI for the customer service number.
The representative who answered the phone knew the exact pickup location, and he even quoted accurate prices, so Rivlin handed over his credit card.
The next morning, he found fraudulent charges on his credit card. He was one of the first victims of AI search poisoning – a growing problem that is only going to get worse.
AI Phone Number Poisoning Is The Newest Threat
Security Researchers from Aurascape have uncovered the latest and most sinister way that AI is being used to scam consumers.
Scammers aren’t hacking the AI systems directly. Instead, they’re flooding the internet with fake customer service numbers on hacked websites, YouTube videos, and Yelp reviews.
ChatGPT sees the scam number on the hacked websites and returns it as the answer to the unsuspecting consumer.
Both Google And Perplexity Got Fooled
The researchers at Aurascape documented the first major AI search-poisoning campaign last month.
When they asked the AI tool Perplexity for “the official Emirates Airlines reservations number,” the AI responded and provided a number, describing it as a hotline for booking and upgrades. The number went to a scam call center.
Google’s AI Overview showed similar problems with AI-poisoned scam numbers.
When asked for Emirates’ U.S. reservation number, it provided fake numbers along with step-by-step booking instructions. The AI cited 10 different “sources” to back up its answer, including what appeared to be government and university websites. All had been hacked or manipulated.
The researchers report that this is not a model generating random numbers; these are attackers quietly rewriting the web that AI systems read and presenting it as reality.
Scammers Target AI Answers At Top Of Google Results
For 20 years, scammers have been gaming Google to rank high on the search results. But with AI agents and phone poisoning, the game has completely changed.
Instead of showing you a list of links to choose from, Google now presents a single answer on top of the results. Scammers now focus on becoming that answer.
They’ve planted fake numbers on government websites, university pages, and popular blogs. On YouTube, channels with names like “Travelsupport-q4n” post low-quality videos, with descriptions stuffed with airline names and scam phone numbers.
And surprisingly, even Yelp can be manipulated with a super simple technique to poison companies’ phone numbers. Fake “reviews” inject fraudulent numbers into what appears to be ordinary customer feedback.
AI systems trust these sources because they come from websites that normally have good reputations
How To Protect Yourself From AI Search Poisoning
There are several things you can do to avoid becoming a victim:
- Never trust AI phone numbers; they are often scam numbers since scammers have poisoned the search results.
- Never call a phone number from an AI assistant without checking it on the company’s official website or app.
- Type web addresses directly rather than clicking links from AI responses.
- Be suspicious of pressure on any customer service call. Scammers create a sense of urgency to stop you from thinking clearly.
- For travelers: Save official airline contact numbers before you fly. When your flight gets canceled, and you’re panicking, the urge to quickly Google a number is exactly what scammers count on.
Thanks for reading.
