AI holiday planning has reached nearly half the people surveyed across five European countries, yet only a small minority were comfortable letting it make the booking. The contrast comes from lastminute.com’s September 2026 Travel Horizons report: 45% of respondents said they had used AI for holiday planning in the previous year, while just 12% were happy for it to book a trip. People appear increasingly interested in letting software organise the options, but far less willing to hand over the final decision.
That is a distinction with real financial consequences. Asking a chatbot to suggest a seaside town or a museum itinerary is a low-stakes experiment. Authorising it to select flights, commit to accommodation and pay a deposit is different. The research suggests that enthusiasm for personalised recommendations has moved ahead of confidence in automated transactions. These are reported attitudes, not proof of how often AI actually completes bookings or how accurately any particular travel agent performs.
AI holiday planning starts with choices, not payments
Holiday planning is an unusually suitable problem for a conversational assistant. A traveller might know their budget, preferred climate and available dates but have no firm destination in mind. They can ask for alternatives, change the requirements and compare activities without starting a new search. The software can help create a shortlist and turn vague wishes into practical questions: how long is the transfer, what is open in the off-season, or which area suits someone travelling with children?
In lastminute.com’s survey, respondents reported using AI for a variety of tasks. Across the five markets, 40% were happy for it to compare prices, 29% to recommend options and 25% to build an itinerary. Those answers involve different tasks and levels of delegation. They should not be read as stages in an actual sales funnel. The same person might welcome automated comparisons while preferring to book directly with a provider they know.
AI can be particularly useful when travellers have competing priorities. A family may want easy transport, accessible accommodation and activities that suit different ages. A couple may be choosing between a lively city break and somewhere quieter. The tool can help surface trade-offs: a cheaper flight that arrives late, an attractive hotel far from the centre, or a package offering convenience at a higher price. A useful answer makes those compromises visible rather than pretending one option is universally best.
The moment money changes hands, trust becomes specific
Booking is not simply the final step of searching. It creates a contract, often with restrictions on cancellation, amendment, luggage, payment and passenger details. One incorrectly typed name can be expensive to correct. A hotel may have an appealing description but a different room category or a substantial deposit. Travellers therefore need more than a recommendation that sounds confident. They need an accurate offer, clear terms and an opportunity to verify the details before making a commitment.
The survey’s 12% figure covers respondents willing to let AI make a booking. It does not measure how many people have tried an autonomous travel agent or had a bad experience with one. The reluctance may arise from unfamiliarity, a preference for personal control or concerns about mistakes and refunds. It would be wrong to turn the number into evidence that automated bookings are currently unsafe. It is better understood as a snapshot of consumer willingness at a point when the technology and its commercial arrangements are still developing.
LiveAIWire’s previous coverage of AI shopping assistants and the hesitation to trust them with checkout describes a similar boundary in retail. Comparing prices is reversible; paying is not always easy to undo. Travel adds another complication: several suppliers may be involved, each with different rules, and a failure in one booking can affect the entire journey.
European travellers do not use AI in identical ways
The report draws on Censuswide research involving 5,001 adults across Britain, Germany, France, Italy and Spain, carried out between 25 August and 1 September 2026. It also uses lastminute.com’s commercial data, which should be kept separate from survey responses. Among the survey’s findings, use of AI for planning was reported by 57% of respondents in Spain and 51% in Italy, compared with 30% in Britain. The figures indicate variation within this sample, not a universal rule about national attitudes.
Age was another dividing line. The report said 73% of Gen Z and 68% of Millennials had used AI for holiday planning, compared with 38% of respondents over 55. These are self-reported measures and broad age categories. They do not establish why any individual adopts or rejects the technology. Experience with digital tools, comfort with experimental services and the type of trip being planned could all influence behaviour, but the survey alone does not isolate their effects.
The company’s own referral data shows that some travellers reach booking websites from AI-powered services. That is evidence of traffic patterns within lastminute.com’s business, not a measure of how many holidays across Europe are sold by AI. Similarly, a higher average basket among referred visitors may reflect who uses the channel or what they search for rather than a direct effect of the technology. The distinction between company data and independent market behaviour is essential when interpreting claims about a revolution in travel.
A confident itinerary still needs ordinary checks
A chatbot may produce a plausible route that contains an inconvenient transfer, an attraction closed on the proposed day or a hotel whose advertised facilities have changed. Such errors are particularly frustrating when a holiday has fixed dates and costs. A suggested restaurant can be replaced; a missed connecting service or incorrectly booked room may be harder to fix. It is sensible to check operating hours, local transport, accessibility and current prices with the relevant providers.
There is also a difference between information about a destination and a purchasable offer. A well-written description of a beach or historic district can be useful long after it was written. Availability, fares and cancellation rules may change rapidly. Travellers should not treat a general AI answer as a live booking confirmation. The final purchase should be completed using current supplier information, with a clear record of what has been bought.
LiveAIWire has examined AI assistants combining shopping and travel planning, where the appeal lies in moving from recommendations towards actions. The same caution applies when a system asks for login credentials, card details or access to messages. Convenience should not require the traveller to surrender more personal information than the task needs. An assistant with permission to act deserves closer scrutiny than one that merely suggests an itinerary.
Who is responsible if an AI-booked trip goes wrong?
Existing consumer protections remain relevant even when software performs part of the journey. The European Commission’s overview of the Package Travel Directive explains that certain combinations of travel services receive particular protections. The rules and rights depend on what has been purchased and how the contract is structured. They cannot be assumed to apply identically to a stand-alone flight, a hotel room and a package assembled through a single seller.
The directive was amended in 2026, with a period for countries to implement changes before the new provisions become binding on operators. That is another reason not to claim an AI assistant provides a specific right without checking the applicable contract and jurisdiction. The traveller still needs to know who the organiser or supplier is, what documentation confirms the booking and where a complaint should go. An automated conversational interface does not replace those details.
A well-designed future booking assistant could provide useful safeguards. It might show a final confirmation screen that names each supplier, highlights non-refundable elements and requires explicit approval of passenger details. It could retain a clear record of decisions and make it easy to correct errors before payment. None of those outcomes follows automatically from conversational technology. They are design choices that providers will have to make and demonstrate.
The sensible compromise is assistance with a human decision
For now, a practical approach is to use AI for inspiration, comparing alternatives and preparing questions, then verify the final details with the businesses that will deliver the trip. Ask a tool to identify possible neighbourhoods, transport options and likely trade-offs. Check availability and prices directly. Read the terms and keep confirmation numbers and contact details somewhere independent of the conversation.
LiveAIWire has previously reported on privacy questions when AI agents handle personal details. Travel bookings involve names, dates of birth, sometimes passport information and payment data, so the cost of careless sharing can be high. The traveller should understand which service receives that information and whether access can be limited or revoked.
The lastminute.com figures do not tell us whether people will always resist autonomous bookings. Familiar products, better safeguards and a record of reliable performance may change attitudes. What they establish is that discovering a trip and committing money to it remain different kinds of trust. The best travel assistant will not merely know where someone wants to go. It will help them understand exactly what they are agreeing to before they go there.
About the Author
Stuart Kerr is Technology Correspondent at LiveAIWire, covering artificial intelligence, cybersecurity and the social impact of emerging technology. LiveAIWire is an independent, human-led technology publication using AI-assisted research, editorial production and original AI-assisted editorial illustrations under his direction.