AI travel advice leaves tourists in trouble at Zimbabwe border

Australians are increasingly turning to artificial intelligence (AI) to plan overseas holidays, but travellers are being warned that relying on chatbots for critical information such as visa requirements can lead to costly mistakes.

Australian traveller Max Phelps discovered the risks firsthand after using an AI chatbot to help plan a multi-country African safari for his family.

The family arrived in Victoria Falls, Zimbabwe, believing they had obtained the correct KAZA visa for their itinerary, which included Zimbabwe, Zambia and Botswana.

However, an immigration officer later informed Phelps that the visa would not cover their planned return to Zimbabwe after they had left the country.

"I landed in Victoria Falls in Zimbabwe and we had the KAZA visa, which we believed was the right visa to get for our trip," Phelps said.

Initially, he suspected the immigration officer was attempting to sell the family an unnecessary visa. He later discovered that the officer was correct.

Phelps said a popular AI chatbot had helped the family optimise their itinerary, identify locally owned safari operators, select flights and plan their travel route.

When asked about visa requirements, the chatbot confidently recommended the KAZA visa and reassured him that it allowed unlimited entry across the participating countries for 30 days.

The family later discovered that this advice did not apply to their particular travel plans.

"It's expired because you've been out of the country for two nights," the border officer reportedly told Phelps.

Fortunately, the problem was resolved relatively quickly. After about 30 minutes and spending around $200 on new visas, the family was able to continue their journey.

The incident highlights broader concerns about travellers relying on generative AI for information that can change frequently or depend on individual circumstances.

Adam Ewart, chief executive of luggage delivery company Send My Bag, said the company's research found that almost 40 percent of Australian travellers were using AI to help plan trips.

"AI is very good at turning complex information into a clear, simple answer," Ewart said. "That convenience is also the danger."

He said AI-generated answers could appear authoritative without clearly identifying the source of the information, when it was last updated or whether exceptions applied to a particular traveller.

"People may also assume AI is checking official government sources in real time when that is not necessarily the case," he said.

Younger travellers are driving much of the adoption, with more than half of Gen Z and Millennials reportedly using AI for travel planning, while older travellers have been slower to embrace the technology.

Ewart warned, however, that being comfortable with AI does not necessarily mean users can recognise when a chatbot has provided incorrect information.

"Being highly confident using AI is not the same as being able to identify when it has made a mistake," he said.

He advised travellers to use AI as a planning assistant rather than an authoritative source for critical travel information.

"Use AI like a well travelled mate: ask it for ideas and use it to prepare, but don't treat it like an airline, embassy or border official," Ewart said.

His warning is particularly relevant for visa requirements, passport rules, border regulations, flight connections and baggage allowances, where an outdated or incorrect answer can result in significant disruption and expense.

"AI can do the homework, but it should never be allowed to mark its own exam," he said.