United has started showing estimated TSA security wait times inside its mobile app at all U.S. hub airports, adding a new planning surface to an app that already handles booking, itinerary management, and day-of-travel logistics. The feature is simple to describe and easy to misunderstand: on the consumer side, it helps travelers decide when to leave for the airport. Underneath, it turns the airline app into a live operational interface that is trying to influence behavior before a traveler ever reaches the checkpoint.
That matters because the value here is not just “more convenience.” It is that the app is moving closer to a decision engine. Instead of presenting static trip information after a booking is made, United is surfacing a prediction that can change how passengers act in real time — whether they head out earlier, cut it closer, or reroute their preflight timing around a busy checkpoint. That is a different product category from a reservation manager. It is software making a judgment about an external system and then asking users to trust it.
Technically, that judgment is hard.
Security queues are one of the messiest kinds of operational data to model because they are local, spiky, and sensitive to conditions that do not behave nicely in aggregate. Staffing changes, weather disruptions, flight banks, holiday traffic, terminal layout, construction, and even a single oversized wave of connecting passengers can produce large swings in wait time over short periods. A station that looks stable in historical averages can become a different problem by 7:15 a.m. on a peak Monday. That is why even a feature that appears straightforward on the screen usually depends on multiple moving parts under the hood: airport-specific data feeds, frequent refresh cycles, a mapping of hubs and checkpoint coverage, and logic for dealing with stale or missing signals.
The Verge’s reporting does not say United is using a specific AI model here, and that distinction matters. In product coverage, “AI” often gets used as a catch-all for any system that produces a prediction, but a forecasting layer is not the same thing as a generative model, and it is not the same thing as marketing copy. The real technical question is whether United is combining historical queue patterns with live airport inputs to produce an estimate that is useful often enough to matter. If that estimate is derived from rules, heuristics, statistical forecasting, or some blend of those methods, the operational test is the same: does it stay calibrated across airports, times of day, and disruption scenarios?
That calibration problem is where the feature gets interesting.
A TSA wait-time estimate is only as strong as the data behind it, and live airport data is notoriously noisy. Some airports have more complete instrumentation than others. Some checkpoints move in bursts that make a single point estimate misleading within minutes. Some terminals have divergent queue behavior depending on whether a bank of flights is boarding or a weather delay has pushed arrivals into the same narrow window. And if the app is updating frequently, latency becomes part of the product: by the time a traveler checks the screen, the queue may already have changed. In a system like this, uncertainty handling is not a footnote. It is the product.
That is especially important because the most obvious failure mode is not that the number is a little off. It is that the number is confidently off at the wrong moment.
If United tells a passenger the checkpoint will take 12 minutes and the real line stretches to 35, the app has not merely failed as a convenience tool; it may have created a false sense of certainty that causes a missed flight, a missed connection, or a last-minute rush that lands on gate agents and airport staff. If it overshoots the wait too often, it has the opposite problem: it trains users to ignore the estimate entirely. In either case, the feature’s usefulness is inseparable from its trustworthiness. A travel app can be wrong sometimes. A travel app that positions itself as the source of airport guidance has to be wrong in ways users can understand.
United’s rollout also reads as a market-positioning move. Airlines have spent years trying to make their mobile apps more than transactional wrappers around tickets and boarding passes. By adding TSA wait times directly into its own app, United is trying to keep travelers inside its ecosystem for a bigger slice of the journey — not just for buying the flight, but for deciding how to navigate the airport. That matters in a competitive environment where the airline app is competing with broader travel tools, airport apps, and generic search products for the role of default interface.
The detail that this is rolling out at U.S. hub airports is important here. Hubs are where timing pressure, connection complexity, and passenger volume make the operational value highest, which also makes them the hardest places to get wrong. If the feature works there, it gives United a credible wedge for expanding the same logic elsewhere. If it does not, the mistakes will be felt in the airline’s most important airports first.
That is why this update is better understood as an operational bet than a consumer perk. United is not just adding a line item to an app; it is asking passengers to use the airline’s software as a guide to a volatile real-world system. The upside is clearer travel planning and a stronger role for the app in the airport experience. The downside is equally clear: once a carrier starts publishing predictions that affect how people move through the day, accuracy, refresh rate, and uncertainty handling stop being backend details and become part of the brand.
