Airlines from Delta to Virgin Atlantic are giving seat pricing to AI that reprices flights in real time, squeezing out the bargains on busy routes while sometimes cutting fares on quiet ones. Old booking tricks lose power against models that weigh dozens of variables continuously — and when inputs include your own search behaviour, it edges toward ‘surveillance pricing’ the FTC is already investigating.
The cheap seat on a busy route is quietly becoming an endangered species. Airlines are handing fare-setting to AI systems that reprice flights in real time, and on popular routes the effect is exactly what you’d fear: the gaps that once let travellers find a bargain are shrinking.
The cheap seat on a busy route is quietly becoming an endangered species. Airlines are handing fare-setting to AI systems that reprice flights in real time, and on popular routes the effect is exactly what you’d fear: the gaps that once let travellers find a bargain are shrinking. Reporting from Bloomberg and The Next Web lays out a shift that will touch anyone who books a flight.
From spreadsheets to real-time repricing
Traditional airline pricing ran on rigid, stepped fare buckets — a limited number of cheap seats, then the next tier, then the next. AI replaces that with models that weigh dozens of variables continuously and adjust prices almost instantly: competitor fares, how you’re searching, historical demand elasticity for that route, weather, local events, the wider economy. Instead of a staircase of preset prices, the fare becomes a moving target recalculated on the fly.
The adoption curve is steep. Delta was reportedly using AI on around 1% of fares in late 2024 and aiming for roughly 20% by the end of 2025, with carriers including Virgin Atlantic also moving in this direction. This isn’t a pilot; it’s becoming standard practice.
Who wins, who loses
The outcome isn’t uniformly bad for travellers — it’s uneven. On high-demand routes, AI mostly means higher fares: the system is very good at identifying exactly how much it can charge before you walk away, and squeezing out the cheap seats that used to slip through. But the same models can cut prices on quieter, low-demand flights to fill empty seats. So off-peak and unpopular routes could actually get cheaper, while the busy flights everyone wants get more expensive and harder to game.
The strategic shift for consumers is that the old tricks — book on a Tuesday, wait for a fare drop, use incognito mode — lose their power against a model recalculating in real time against your own behaviour. Flexibility (on dates, routes and times) becomes the main lever travellers have left.
The “surveillance pricing” worry
The sharper concern is what the models are pricing against. When the inputs include an individual’s search behaviour and data profile, dynamic pricing edges toward what critics call “surveillance pricing” — charging different people different prices based on what the system infers about their willingness to pay. That’s a meaningfully different thing from adjusting prices to overall supply and demand. The US Federal Trade Commission has already opened a civil investigation into whether airlines are using individualised data profiles to raise the prices specific consumers pay.
That distinction is the whole ballgame. Pricing a route by aggregate demand is ordinary revenue management. Pricing you by what your data suggests you’ll personally tolerate is something regulators — and passengers — are right to examine closely.
The bigger pattern
Airlines are an early, visible case of a broader move: AI turning static, published prices into personalised, continuously-adjusted ones across industries. What plays out in air travel — real-time optimisation, uneven consumer outcomes, and a regulatory fight over individualised pricing — is a preview of arguments coming to hotels, event tickets, ride-hailing and beyond. The cheap seat isn’t just getting rarer; it’s becoming a test case for how much a company is allowed to know about you before it sets your price.
Sources: Bloomberg; The Next Web.
