Are Airlines Liable When Their Chatbots Give Passengers Wrong Travel Advice in 2026?

Direct Answer: Yes, Airlines Can Be Liable for Misleading Chatbot Advice

Yes, an airline can be held responsible when its chatbot gives a customer inaccurate information about a refund, cancellation right, baggage allowance, compensation, or other booked service. The central legal question is not whether artificial intelligence made the mistake; it is whether the airline supplied, controlled, and put the chatbot into service as part of its own customer-support system. In the 2024 Moffatt v. Air Canada dispute, British Columbia’s Civil Resolution Tribunal held that Air Canada could not distance itself from misinformation published by its website chatbot. The airline was ordered to compensate the passenger for the difference between the bereavement-fare credit he had been promised and the lower price he had actually paid, together with the relevant fees.

Also worth reading: What is the agentic AI travel compliance checklist and how can airlines and OTAs implement it in 2026? · What are the NDC offer reprice flow best practices for travel agencies and airlines? · NDC vs GDS distribution: which channel should airlines and travel sellers actually use in 2026?

That decision is unusually clear for a chatbot case, but it did not create a universal rule under which every AI answer automatically becomes a binding airline promise worldwide. Liability still depends on applicable consumer law, contract terms, the airline’s responsibility for the system, the exact representation made, and whether the customer relied on it to their detriment. A chatbot hosted by an unrelated third party may produce different legal questions from a chatbot integrated into the airline’s website, app, booking flow, or telephone support operation. Even so, a passenger should not be expected to investigate prompt engineering, model architecture, or vendor contracts before challenging a plainly false statement made through an official airline channel.

The most important practical rule is simple: preserve the conversation and act quickly. Do not rely only on a screenshot if the chatbot can delete or replace messages; export the full transcript, record the page and account details, and keep the booking confirmation, fare rules, payment records, and correspondence. A claim is harder to assess when the passenger cannot identify the exact words, date, booking reference, or promised remedy.

Why the Air Canada Chatbot Case Changed the Conversation

In Moffatt v. Air Canada, passenger Jake Moffatt asked the airline’s chatbot about the bereavement-fare policy for a flight from Vancouver to Toronto. The chatbot told him that he could apply for a refund of the difference between the bereavement fare and the price he had paid, provided he made the request within 90 days. Air Canada later denied the request and removed the relevant wording from its website, arguing that the chatbot was responsible for the information and that the airline was not bound by it. The Civil Resolution Tribunal rejected that position. The decision was issued on 14 February 2024, and the compensation order is widely reported as CA$650.50 in damages, with additional fees and expenses forming part of the total amount paid.

The reason the case attracted wide attention was the contrast between the airline’s legal position and the ordinary customer experience. Website advice can reasonably appear authoritative because it is presented by the company selling the flight, appears on its official domain, and responds in a polished conversational format. The tribunal also criticized Air Canada’s attempt to suggest that it should not have to explain how its website was built. In effect, the company could not treat the chatbot as a dependable source of instructions when that helped customers, while treating it as an independent and unaccountable actor when an instruction proved wrong.

The award was modest, but the decision’s public importance was greater than its monetary value. The reports note a dispute involving roughly CA$880 when accompanying amounts are considered, while the core damages commonly reported from the order are CA$650.50. The case therefore was not about obtaining an unusually large payout. Its significance came from confirming that companies may remain responsible for information supplied through AI-enabled customer interfaces.

How Airline Chatbot Claims Are Evaluated

A court or tribunal will usually examine several connected elements rather than asking only whether the chatbot was technically correct. The first question is whether the airline controlled the relevant channel. A chatbot on the airline’s domain, embedded in its app, linked from its booking confirmation, or used during a support handoff is more closely connected to the carrier than a generic public tool. The second is whether the statement was clear enough to influence a reasonable decision. Definitive wording such as “you are entitled to a refund within 90 days” is more consequential than a vague or obviously automated answer that never resolved the passenger’s question.

The claimant must also show detrimental reliance, meaning that the false or misleading statement caused a real loss or prevented them from obtaining a benefit they would otherwise have pursued. Common examples include booking without travel insurance because the chatbot said none was needed, buying a fare the passenger believed was refundable, failing to make a cancellation request within a real deadline because the chatbot supplied the wrong date, or paying for a service the passenger believed was included. A claim based solely on dislike of the answer, without financial or contractual effect, may be weaker. The evidence should connect the chatbot statement to the passenger’s decision, ideally through messages sent immediately afterward.

Contract wording, consumer statutes, and the carrier’s published policies then determine what remedy may be available. The chatbot cannot override every term in a ticket contract, but a term may itself be challenged if it is hidden, unfair, or inconsistent with mandatory consumer protection. The airline’s conduct, refusal to correct the error, and response after discovery can also affect the outcome. Although AI is the delivery mechanism, the legal object is usually the airline’s representation and the consequences of relying on it.

What Passengers Should Do When Chatbot Advice Is Wrong

The first practical step is to capture evidence while access is still available. Save the full conversation, including the question, answer, timestamps, booking reference, and any buttons that allowed the passenger to continue. Take screenshots showing the airline’s logo, domain, app, or support identity, but do not treat a cropped image as the only record. A screen recording can help demonstrate the route used and any chatbot disclosure; however, screen recordings do not replace a text transcript when the underlying message is technically accessible.

Next, compare the statement with the fare rules, conditions of carriage, refund policy, and booking confirmation. Highlight any direct conflict, but do not argue that the chatbot overrides a clearly stated term unless the airline represented that it would. Send a concise complaint through the airline’s formal customer-service or feedback channel, identifying the booking reference, exact chatbot quotation, date, and requested remedy. Ask for the information to be placed in writing and for the record to be retained. This is not a requirement for filing a claim, but it can show that the passenger acted promptly and treated the issue seriously.

If the airline refuses to resolve the matter, preserve every ticket number, boarding pass, payment receipt, and written response. A passholder may be dealing with both a fare dispute and a cancellation or delay claim, so the chatbot evidence should be organized alongside those records. A small-claims tribunal, ombudsman, consumer-protection agency, payment provider, insurer, or court may be relevant depending on location. The passenger should select the route appropriate to the jurisdiction rather than assuming that the Air Canada decision applies automatically as binding precedent everywhere.

Comparing Remedies and Alternative Support Channels

Choosing a remedy depends on the error, the passenger’s loss, urgency, and jurisdiction. A new chatbot answer may correct future guidance but does not itself pay compensation. A complaint to the airline is useful, yet it can be slow when the dispute concerns a departing flight. Formal adjudication can produce a decision, but filing costs, evidence, travel, and procedural rules can make it disproportionate for a small loss. Alternative dispute resolution and the airline’s customer-feedback process may be faster, although they are not always neutral or binding.

A written human-agent escalation is often the most useful alternative to repeating the same question to the bot. The passenger can paste the erroneous answer and ask the agent to explain which term or policy controls. That comparison is valuable because a human correction does not automatically prove the earlier statement was binding, but it can establish the carrier’s later position. If no one disputes the facts and the loss is small, a direct refund request may be more efficient than litigation.

Common Mistakes That Can Weaken an Airline Chatbot Claim

A frequent mistake is relying on a short screenshot without preserving the URL, time, and complete exchange. Another is describing the chatbot as if it “lied,” which is emotionally understandable but legally imprecise. A strong claim focuses on the exact representation, the airline’s control of the channel, the passenger’s reasonable reliance, and the resulting loss. Calling the system a liar may distract from evidence that directly answers those issues.

Passengers also make the mistake of overlooking an earlier deadline. A chatbot may mention a 24-hour cancellation window, a 90-day refund request, or a requirement to contact the airline before departure, but the wrong date or refund amount can still matter. If a real deadline is approaching, the passenger should submit the available request immediately and obtain confirmation, even while disputing the chatbot’s interpretation. Waiting to perfect a complaint could cause loss of rights that were separate from the eventual compensation claim.

Another error is assuming that a binding fare rule defeats every chatbot claim. Rules can help the airline show what the contract required, but consumer regulators may examine misleading explanations, unfair exclusions, and representations made during booking. Conversely, assuming that the Moffatt case guarantees a refund in every country is also mistaken because the decision is not necessarily binding beyond its forum. Finally, deleting messages, using an edited transcript, or exaggerating the financial loss can damage credibility. The passenger should preserve the original material and calculate the claimed amount from receipts rather than an estimate.

When to Act, and What It May Cost

Time is important even when the full amount of the refund cannot be recovered. A passenger who learns that chatbot advice was wrong should preserve the evidence on the day the discrepancy is discovered, send a written complaint promptly, and avoid missing any stated booking, cancellation, or complaint deadline. As a general travel reference, many airlines require changes or cancellations through their official channels and may impose fees unless the fare is demonstrably flexible. Those fees are contractual or policy costs, not automatically compensation, but they should be documented if the passenger claims the chatbot caused the transaction.

A formal claim can be inexpensive in some consumer or small-claims systems, while other routes involve application fees, legal representation, travel, or loss of cabin baggage if a hearing is remote and the passenger cannot attend. A passenger seeking only an airline correction can usually use ordinary customer service at no extra charge. A new booking made after discovering the chatbot error may create a larger loss than the original dispute, so the passenger should calculate the difference between the actual purchase price, any refund, and the price of the replacement trip.

Compensation for denied boarding, cancellation, and delay is separate from a claim about chatbot misinformation. Depending on the country, passenger rights may include rerouting, care, meals, accommodation, or refunds under laws such as European Union Regulation 261/2004, the Montreal Convention, or local consumer statutes. Those systems often have their own deadlines, including short periods for reporting disruption in some circumstances. A passenger should therefore pursue the disruption remedy and the chatbot-evidence dispute in parallel when both are relevant, rather than assuming one claim automatically covers the other.

The 2026 View for Travellers and Airlines

By September 2026, the safest practical assumption is that an official airline chatbot is a communication channel, not a legally isolated experiment. The Moffatt decision remains a prominent warning that a carrier may be unable to shift responsibility to the chatbot operator when the system is part of its customer-facing service. Airlines have responded by restricting what their bots can say, directing users to policy pages, adding disclaimers, and routing complicated requests to human agents. Those changes may reduce the volume of advice, but they do not automatically remove responsibility for statements the bot still gives.

For travellers, the distinction between legal advice and operational guidance remains important. A chatbot can help locate a policy or summarize a fare condition, yet its answer may be incomplete or incorrect. Passengers should read the controlling fare rules, compare the response with the booking, and ask a human to confirm any decision involving a large expense, a non-refundable ticket, medical travel, a bereavement fare, baggage, or a missed connection. When the answer conflicts with a clear policy, keep both records and dispute the discrepancy rather than silently accepting the more expensive outcome.

For airlines, the case is best understood as an accountability and recordkeeping issue. A company that deploys AI should know which answers it permits, monitor errors, preserve logs, train staff to handle escalations, and correct inaccurate information promptly. The CA$650.50 core award in the Air Canada case was not large enough to create a universal liability formula, but the reputational and regulatory attention surrounding a small-claim dispute is substantial. The practical lesson is not that airlines must accept every chatbot answer; it is that they cannot safely present an official bot as authoritative while refusing accountability when that authority produces harm.