United And Delta Have Officially Banned Humanoid Robots From Flying As Passengers

United and Delta Air Lines have followed Southwest in banning humanoid and animal-like robots both from flying as passengers and also as checked baggage. Live and Let’s Fly reported an internal United bulletin that flagged the carrier’s policy took effect July 23.

“Starting July 23, humanoid and animal-like robots are not permitted in the cabin… [We will] continue to evaluate if and how we can transport them safely.”

United defines a humanoid as an AI-driven or remotely controlled machine that mimics how a person looks, moves, thinks or speaks. An animal-like robot is one that looks or behaves like an animal. Small robot toys and gadgets may still travel if they meet ordinary carry-on and battery rules.

Employees have reportedly been told to stop these robots at check-in and boarding. If one reaches the aircraft, flight attendants are supposed to contact the gate and arrange to have it removed. This is also now reflected on United’s public dangerous-items page.

Delta Has Quietly Done The Same Thing

Delta’s published baggage policy says it has temporarily banned any robot designed to resemble or imitate a human or animal in appearance, size, movement or behavior, whether in the cabin or checked. Human- and animal-like toys are exempt if they fit normal carry-on limits, can be stowed properly and comply with battery restrictions.

Delta also says these robots are not assistive devices and are not trained service animals. That distinction works for today’s event-rental robots, though it’s going to become much harder to maintain as robots become personal aides for older people and travelers with disabilities.

That leaves American as the only one of the four largest U.S. carriers without a published humanoid prohibition. Of course, ordinary battery, size, weight, cabin-baggage and cargo rules still apply.

Southwest Adopted A Rule After Robots Actually Showed Up

I first wrote about this in May, before the later wave of coverage. A passenger bought an extra seat on Southwest flight 1568 from Oakland to San Diego for Bebop, a four-foot, roughly 75-pound robot. Crew first objected to its aisle seat and then to its battery. Bebop was moved to the window, the battery was removed, and the flight eventually left about an hour late.

Then a 3.5-foot robot named Stewie flew from Las Vegas to Dallas. Its owner bought the extra seat used for fragile cabin-seat baggage and fitted the robot with a smaller battery comparable to one in a laptop. Stewie walked through the terminal, boarded and sat at the window. Two days later, Southwest issued a companywide safety alert barring human- and animal-like robots regardless of size or purpose.

Batteries And Projectiles Are Real Problems

FAA rules generally allow lithium-ion batteries up to 100 watt-hours. Batteries from 101 through 160 Wh require airline approval, with limits on the number carried. Above 160 Wh they are not permitted on a passenger aircraft under the ordinary personal-electronics exception. Spare batteries must remain in the cabin, where crew can detect and respond to a fire.

Lithium cells can enter thermal runaway after damage, overheating, overcharging, improper packing or a manufacturing defect. It can happen without warning.

The 46.8-volt, 9 amp-hour pack documented for the Unitree G1 is about 421 Wh, more than twice the passenger limit. Figure’s current humanoid battery is 2.3 kWh—more than fourteen times the limit and is integrated into the torso as a structural part of the robot.

Southwest banned Stewie even after it was fitted with a compliant smaller battery, though. United’s rule applies regardless of size or purpose. A human-shaped shell with no battery is prohibited while a box-shaped machine with a permissible lithium battery may be allowed.

There are non-battery issues, too. A 75-pound machine can become a projectile. Motors could activate accidentally. And cabin baggage has to be secured without exceeding seat or floor limits or blocking an aisle or exit.

An interim ban can make sense while airlines develop a process. A robot traveling in the cabin should be secured as approved cabin-seat baggage at a window in a non-exit row, and it should not walk around during the flight. Batteries over the passenger limit shouldn’t travel as a standard carry-on.

Our Travel Reality Is About To Change

Delta’s policy declares that these robots are not assistive devices. But the federal definition is functional, not based on shape: an assistive device is equipment that helps a passenger with a disability hear, see, communicate, maneuver or perform another function of daily life.

A future robot that guides a blind traveler, lifts bags for someone with limited mobility, monitors an older passenger’s health or helps someone communicate will fit that definition. It would not be a service animal (under current rules that means a trained dog) but it could be assistive equipment. Airlines have obligations to carry assistive devices when they can be stowed and transported safely, though hazardous-material rules still apply.

The idea that “a humanoid robot can never be an assistive device” is simply not going to be reality soon.

Airlines Are Already Putting Humanoid Robots To Work

The irony is that airlines are not merely waiting for this future. Japan Airlines began a multi-year trial at Tokyo Haneda this spring using humanoid robots for ground handling. It is starting with cargo-container movement and studying baggage and cargo loading, cabin cleaning and eventually the operation of ground-support equipment.

JAL explains exactly why the robots are humanoid: airports and aircraft were built around human movement. A human-shaped machine may be able to use the existing facilities and equipment without an airport first rebuilding everything for specialized automation. Airbus is also in an early test phase with a humanoid robot for aircraft manufacturing.

And Aeroflot low-cost carrier Pobeda already put “Volodya,” a Unitree G1, on a regular Boeing 737 flight from Ulyanovsk to Moscow. It greeted customers, made limited safety-demo gestures, walked the aisle and sat down for takeoff and landing. Human flight attendants still performed all required service and safety duties.

The Faster Disruption Is Already Inside United

Physical robots are a little farther off than the disruption from large language models. But United has already said AI is helping it shrink management staffing. Its CFO said management headcount was 4% lower than the year before and that the airline planned another 4% reduction in 2026 as process changes and AI made headquarters work more efficient.

The first airline jobs to change are not likely to be flight attendants and pilots. They are contact-center work, back-office processing, marketing and loyalty servicing, revenue management, network planning, crew pairing and disruption recovery, and maintenance planning. Much of this is well-paid white-collar work. Airlines are heavily regulated, so cockpit and cabin-crew mandates will probably make those jobs among the last to change even when the technology is capable.

Eventually, though, United may be banning a robot in the passenger cabin while relying on machine intelligence up front. A robot pilot! And a ban on machines will be strange when robots are building aircraft, moving bags, and cleaning cabins.

We are too status quo-biased in how we think about this. We see today’s awkward robot and assume tomorrow will look like today with a slightly less awkward robot. We look at 2023’s unreliable chatbot and miss how quickly it is improving or how much repetitive professional work it can already take over. United is right to say it may revisit the rule.

About Gary Leff

Gary Leff is one of the foremost experts in the field of miles, points, and frequent business travel - a topic he has covered since 2002. Co-founder of frequent flyer community InsideFlyer.com, emcee of the Freddie Awards, and named one of the "World's Top Travel Experts" by Conde' Nast Traveler (2010-Present) Gary has been a guest on most major news media, profiled in several top print publications, and published broadly on the topic of consumer loyalty. More About Gary »

More articles by Gary Leff »

Comments

  1. confusing what airlines do with AI or robots in their HDQ or even terminals vs. in aircraft cabins is the issue.

    Even beyond the battery issue, robots have yet to prove that they can assist in an emergency including evacuate itself following crew commands on a level comparable to a human, even considering that all humans are not at the same level in following directions or assisting other people.

    and unlike trained dogs which can demonstrate that they can assist a human to board and use the lav during the flight, robots cannot be labeled as such. There might be a standard for labeling a robot as an assistive companion but it does not exist now to the same standards that can be applied to trained dogs

    OTOH, robots won’t even leave their calling cards but probably won’t sit at your feet on the floor

  2. Robots will block aisles and rows in an emergency. And the safety issue with batteries is real.

  3. Trust me this will change in next few years. Not only will humanoid robots become more common (I plan to buy one myself) but the ADA may prohibit restrictions just like they do today with service animals.

  4. Domo arigato, Mr. Roboto.

    When Matthew over at LALF posted on this, he focused on the lithium-ion battery fire risks; Gary, appreciate you diving deeper into the FAA’s actual technical and watt-hour limits. Riveting stuff.

    @Tim Dunn — Sure seems like the primary use of AI today is leveraging hype to juice stock numbers, lay off customer service staff, and deceptively micro-target consumers so they pay more than they should. Over at OMAAT, I believe it’s the resident troll Alert who’s always pushing to replace human pilots and ATC with algorithms. You cool with that, or would you still prefer at least two humans up-front?

    For fun, I recommend watching *The Creator* (2023). It takes place in 2070 after the Western world has banned AI, while “New Asia” has fully embraced humanoid simulants and granted them equal rights. Following that timeline, we’d probably have to false-flag nuke ourselves first before a true ‘robot rights’ movement can emerge in the cabin… (calm down, it’s just science fiction…)

  5. United defines a humanoid as an AI-driven or remotely controlled machine that mimics how a person looks, moves, thinks or speaks.

    We see a lot of those at airports already. People running on mental autopilot via their “remotely controlled machine” screens.

  6. @Denver Refugee — Ah, reminds me of that Bo Burnham song… ‘Welcome to the Internet, have a look around, anything that brain of yours can think of can be found, we’ve got mountains of content, some better, some worse… it was always the plan, to put the world in your hand… apathy’s a tragedy, and boredom is a crime, anything and everything, all of the time.’

  7. Really great. Now we’re going to have Humanoid Robots taking up multiple seats in lounges and shoveling all the food in their compartments for later.

    And don’t even get me started with all their unruly drone children knocking things over and running into people.

  8. ( d’ Eskimo) is strong on replacing humans , the other is incubating with nonsense …waiting . I need troll aid.

  9. 1990
    first, I think the market overall is being a little more circumspect w/ AI and its valuations than they have been with other “big ideas” the market has been fixated with.
    Companies want to use AI to reduce costs; I am sick and tired of AI powered voice response units that give partial answers and then dump the call when you keep asking the question that they can’t figure out how to answer and won’t transfer your call to a human.

    as for planes, I don’t see less than 2 pilots in the cockpit during critical parts of flights but I do think it is possible to change the staffing on ultra long haul and long haul flights with better automation on the ground and in the air – but I am not sure I will still be around by the time all of those “ifs” are satisfied.

  10. @Maryland — It’s fascinating to see who else jumps around between here, there, LALF, Cranky, FM, DoC, etc. Those are the five I’ve noticed to be most active, outside of literal Reddit, BlueSky, etc. I see you mostly stick to the first three. I recall Gary used to do a comparison of site traffic among the blogs. Wonder if he’ll do another post about that soon enough.

  11. OK, so the more “human-like” the robot, the more it gives me the creeps…HOWEVER, once the robot is in a NON-emergency window seat, with its seatbelt fastened throughout the entire flight (I mean, it’s not like it’s going to have to get up to use the lav or need to vomit after eating a contaminated salad), I see no reason why it cannot fly…presuming a) their seat is paid for, b) they can get through the TSA screening and show proper ID, and most importantly, c) the battery issue is resolved.

    Robotic animals are another story. Do they fly in the hold in their own kennel? Do they sit at the feet of their owner? Must the owner fly in the middle seat so the robotic dog can have the window? All I can think of now is that scene from “Sleeper” where Woody Allen is given a robotic dog (“Rags”) as a pet and he asks, “Is he housebroken, or is he going to leave batteries all over the floor?”

  12. @Tim Dunn — When it comes to AI, I’m a little more concerned these days that private credit is acting as a ‘shadow bank,’ and because those deals on data centers, etc., are largely hidden, we won’t know when the ‘squeeze’ will hit. Could a modest rise in interest rates starts a chain-reaction? If those loan costs increase and developers default, then those private funds evaporate. Oof, gee boy howdy, then would the super-rich might start to ‘feel the pain’ like the rest of us have been.

Leave a Reply

Your email address will not be published. Required fields are marked *