Consumers trust AI to find deals, but not to spend their money

Users are comfortable with AI searching and comparing options, but draw the line at letting it book or buy things on their behalf. Some companies say it’s for the best.

United Airlines announced it will block third-party AI agents from accessing its ticket-purchasing or booking systems on behalf of travelers.

The reasoning behind this decision is to ensure the safety of users’ data.

“Granting that level access has the potential to expose personal information or result in bookings, account changes, or use of MileagePlus miles without your approval,” United Airlines said.

The company noted that being checked in to a flight by a third-party AI agent may violate federal regulations. For example, “flight check-in requires human acknowledgment that bags do not contain banned substances like explosives,” said United Airlines.

Going against the AI agent trend?

United Airlines’ decision to suspend AI agents from assisting customers may seem to run counter to the growing trend of using AI agents to optimize everyday life, from sending summaries and setting reminders to booking reservations and making purchases.

It was recently reported that the short-term rental platform Airbnb is developing an AI travel agent, set to be available in 2027.

The agent would help the platform’s customers to research, plan, and book their trips.

While it’s not yet clear how much of the booking process the agent will handle, it is expected to help users plan trips, invite other travelers, and resolve issues involving reservations, refunds, and rebookings.

Despite the company’s future plans, recent trends in using AI agents to order food, book a stay, or get a ride suggest that customers aren’t particularly keen on handing these tasks to AI.

Of the roughly 500,000 bookings Airbnb receives each day, only a very small fraction is made by AI agents such as Muse, according to Brian Chesky, the company’s CEO.

Uber witnessed the same trend: around 320 million trips and orders were received via Uber and Uber Eats, while only 4 rides and 500 food orders were placed by the AI agents, TechSpot reported.

Good as an assistant, not trustworthy enough as a shopper

49% of consumers reported using AI when shopping, according to last year’s survey by the agency PartnerCentric.

Of over 1,000 people, 83% see it as an assistant, with more than half viewing it as a tool for finding the best price.

While the numbers suggest that users trust AI to find the best deals, provide convenient comparisons, or speed up product searches, that trust diminishes when it comes to making a purchase.

Only 3% of respondents see an AI agent making a purchase on their behalf.

A similar trend is evident in research by Boston Consulting Group (BCG), which found that 80% of US consumers reported using AI.

96% of them use AI for product search, 80% for comparison, and 24% for purchasing, with only 4% of users saying they would allow an AI agent to make a purchase for them.

What’s the worst that could happen?

A number of banking institutions, including Bank of America and NatWest, have recently warned users that giving AI agents access to their banking details could lead to reckless spending or money loss to scams and fraud, Reuters reported.

Last year, an influencer, Ariana Patiño, booked a stay at a luxury hotel in Las Vegas, only to learn that the place she paid to stay in was AI-generated.

AI also led an elderly tourist couple to travel in Malaysia, covering 300 kilometers (186 miles) to visit a cable car they’d seen in an online video. After arrival, they learned that such an attraction doesn’t exist.

Considering such and similar situations, where users were bamboozled by AI hallucinations, providing it with sensitive information, such as one’s banking details, could have even worse consequences.