Cops won’t need a warrant to pull the data off an autonomous car
arstechnica.com
arstechnica.com
As engineers, we have an obligation to think through privacy concerns.
These (and nearly all) devices should either not retain data or should encrypt it. Also my self-driving car better not be phoning home or getting updates over the air. I've seen what happens in Fast and the Furious 8… https://fsmedia.imgix.net/19/8a/19/0d/3e81/44e8/ab9d/679cbb2...
This one's tricky...
I work for a semi-governmental organization in the Netherlands and we track ships. Tracking company-owned commercial ships is no problem, but we are heavily restricted by law in what we can do with the location data of privately owned ships because people live on them.
If you go to a public concert, people can look at you, note that you were here and tell that to others. That doesn't violate privacy because you had no privacy there to begin with. And in a similar manner, if you drove last thursday on a particular road, whoever bothered to look at you and your car is allowed to do so, note that you were there, write it down and tell that to others.
Privacy is about your private stuff - what you do in your private space, what you have in your private items. If you go out in the public, the things that you do (and where, and when, and with whom) are not private anymore.
The court argued that a car is so public and visible that the GPS tracker is only collecting "public knowledge" anyway.
[0] https://www.tagesanzeiger.ch/panorama/vermischtes/wenn-am-au...
We specifically use AIS [1], tracking ships is quite easy because of it. Other methods are radar.
Because it so relatively easy, there are a lot of commercial companies doing it too. The most well-known is probably MarineTraffic [2]
[1] https://en.wikipedia.org/wiki/Automatic_identification_syste...
Enforcing an prohibited zone, which exists for safety, probably is well within parameters. If you only track and save the vessel that are very close to your drilling platform because you can't track a persons movements from just that data.
I am not sure where you believe "no one" is asking for, or collecting historical data. In fact everyone from the Manufacturers, Potential Advertisers, Law Enforcement, everyone is in fact asking for and getting historical data
so where where are you getting this myth that no one is
FAA and NTSB agencies aren't known for abuses of power. Local law enforcement, on the other hand, routinely abuse power (civil forfeiture, unwarranted traffic stops, racial profiling, police brutality, falsifying evidence, lying under oath, illegal search and seizure...).
Self driving cars should reduce accidents, because computers > monkey brains, so granting the police unrestricted access to data shouldn't impact safety improvements.
I guess here is the real problem. I think it's better to go after the real problem instead of trying to work around it.
Or, say, forced labor for the entire population. If people experience negative feelings while imprisoned, then maybe we should work on the real problem of those feelings, not challenge the nice neutral forced labor concept.
My point is that power over ordinary people generally corrupts, and the only real solution is to reduce and distribute that power so that individuals have more autonomy. Mass surveillance from AVs is a direct threat to your human autonomy.
Either let us watch the watchers, or don't let them watch.
That suggests one possibility would be to separate law enforcement agencies responsible for the management of automotive violations from law enforcement agencies for everything else.
Highway patrol + DMV enforcement arm for registration, safety, and moving violations. Normal police only have jurisdiction for vehicle stops for things like stolen vehicles or reckless endangerment.
Might solve a bunch of other civil liberty issues.
You start off on the assumption that everyone agrees that aircraft should be tracked, and that data should be open to law enforcement review with no warrant. To the extent the data can be justified for safety it should be limited to that purpose only and barred by law from being used in any other manner. That is not how the law is today, as these records are used in all kinds of legal cases (both civil and criminal) outside of need for safety.
People that are actually pro privacy understand how these records are used and do not desire to see it expand to the tracking of every car on the road as a car is much more personally identifiable than a plane, and is in much more common usage.
in short, Maybe automobiles should not be treated differently and maybe it is time to look at how aircraft are tracked and what that tracking is used for changing the law to be more privacy focused
Do you want a gold star? that in no way proves you have a "vested interest" in privacy, that just proves you have give some money to charity.
> The point is there is a strong and historic precedent that transportation systems are monitored
Yes and people that have a "vested interest" in privacy are concerned about and have problems with the current levels of surveillance on these systems,
You do realize there is a difference between public and private transportation, right?
I have yet to see a precedent that supports the government getting as much data as TFA talks about while using my private transportation system. Or the Supreme Court deciding that making a person wait until "cops with bomb and drug sniffing dogs" show up to a traffic stop without probable cause is an "unreasonable search". I could go on and on...
That's a completely different technical problem than having the vehicle store some period of data.
I guess authoritarians will push to have the ability to install monitoring programs locally.
Body camera already demonstrate that (some) police will actively disable recording devices.
https://www.thedailybeast.com/baltimore-cops-turned-off-body...
One department tried to claim that their cameras were "malfunctioning" in the majority of FOI / criminal cases where they were requested.
We allow you to drive on the public roads, in exchange for you giving up your driving privacy. This is why cops can stop you and give you a breathalyzer at will, because that's the trade-off you agreed to when you got your driver's license.
If you want to encrypt your driving data, you should only drive on private roads.
I never assume that driving is a right, instead of a privilege, and the best way to maintain privacy is to just not participate in public activities in the first place..
Regarding public roads, we need licensing and insurance because we can severely damage other people sharing the same space, but that's not necessarily linked to consent to eliminate privacy there. We expect as reasonable for license plates to be recorded when something bad or suspicious happens, and that things like comprehensive license plate tracking is an invasion of privacy.
You can extend that reasoning to public sidewalks. I mean, walking is indeed a "public activity" that takes place in a public space, so I assume that following your line of logic you'd be ok with any public authority following you or me (or anyone else for that matter) without any warrant, wouldn't you?
Uh-huh...
It's been a long time so perhaps I missed the part where I signed away my constitutional protections in exchange for my privilege to operate a motorized vehicle.
Unless by "at will" you mean "under reasonable suspicion" and/or "with probable cause"?
Yet again you people make me sad...
Push back e.g. updated model data to help with updating routes, real-time construction developments, new driveways, ways humans make mistakes on the road, etc? I believe that every AV company will be doing this, and hopefully this data is useless to law enforcement.
OTA updates? Tesla already does. It won’t even be a choice, and that’s probably a good thing. Many people who don’t work with computers professionally don’t see the importance of updating things. In a situation where software updates could literally prevent deaths, I’m pretty sure that it will be mandatory to have vehicles receive OTA updates that can’t be disabled.
By far the simplest way to prevent this vulnerability is to prevent automatic connections to the outside world.
Consider the following scenario:
Driver A and driver B are involved in an accident. Driver C is in an autonomous vehicle that "witnesses" the accident, by being in close proximity to the event.
If driver A or B have the same insurance company as C, that company could get the data from C to determine fault in the accident that C's vehicle witnessed... or better yet, show that both drivers were at fault and they will not pay anyone anything.
As the number of autonomous vehicles rise, the amount of data available to companies will grow, too.
I would be surprised if there isn't language in the policies being underwritten today that would make such a thing possible... possibly even mandatory.
I think I'm misunderstanding, but when you say "make such a thing possible", what exactly are you referring to? The scenarios in your post I would think are heavily dependent on the considerations agreed upon in the policy between insurer and insured, so while Driver A may have a policy that sets limits on liability, Driver B may have a completely different set of considerations in the event of vehicle collision, what Driver C's car "sees" would just be supplementary, wouldn't it?
There's probably reason to make sure that policies don't contain weird provisions related to that data collection, but there isn't really a strong case for hiding driver information, there is enough variation in drivers to justify varied premiums.
Given each AV has complete 360 degrees camera coverage (for detection of other vehicles, if Tesla, for debugging, if others which rely primarily on Lidar) and this camera data is very valuable for debugging and as a liability reducer, this data is very probably to be kept, then transferred to a datacenter. So each AV is also a bunch of mobile security cameras.
LE officers can access those recordings with the same ease they access fixed security cameras, once an investigation starts. Very easily, even without a warrant.
One increased difficulty is the need to ask AV makers which vehicles were near the incident at the time, so as to know which AVs camera recordings to ask for. That location data is protected, at least in some US states.
For development, sure, in test fleets, sure, but production cars won't bother with that.
So you wouldn't record and store an officer's complete shift. When the officer turned on his lights or siren, the device would rewind 30-45 seconds and start recording. It would then record until the lights/siren were disengaged. That is the video that would be uploaded.
So I think a more likely scenario is that the autonomous cars would in fact permanently store and share all GPS data, but data from the sensors would only store what was recorded with a reasonable buffer around a fault event such as a collision.
Also don’t forget Google’s substantial investments in both AVs and ad-surveillance. They have a strong incentive to make cars phone home the whole time, with as much as possible. And so will their lobbyists.
I would prefer such a world. It provides a different legal context where law enforcement would need to compel each individual owner to turn over data. Such compulsion could certainly still be realized, but it's a more significant hurdle than forcing the manufacturer rather than the user, and it's more likely to see challenges.
1. Means people have to build and test two systems
2. It's confusing to the normal user
3. It's better for the product if it's shared as they can iterate and improve based on the data.
This actually applies to loads of products, like voice recognition (Google has recorded and stored everything you've ever said to 'ok Google's on their servers).
So I don't think we'll see a solution until we see a massive shift in privacy becoming a key election issue. Which it probably won't for decades or until a ruthless dictator takes over a developed country (sort of happening in Turkey).
With that being said, privacy concious people should have the ability to opt out, and either not have their data stored permanently at all, or allow them to store it on a personal server. If you chose to do that, it's on you to make sure your server is setup in a way to ingest the data, so no additional work for the vendor.
I suspect in reality most vendors would not provide that option, because data is so valuable, which is why it's important to have laws that allow us to protect it.
Also considering that all the big car manufacturers are working on business models where usually driver and owner are different entities this even gets more involved.
We had an interesting case in Germany where data ended up in the court room neither the manufacturer nor the car-sharing provider claimed to have.
A man was sentenced to 33 months in prison for vehicular manslaughter. The verdict was based on location data obtained by the court.
Both, the manufacturer (BMW) and the provider (DriveNow), maintain that they don't collect personally identifiable location data. The location profile of the convicted man was created by the court from the data it received from both companies independently.
[1] http://www.sueddeutsche.de/auto/datenerfassung-im-fahrzeug-d...
OT: Original source is in German and I wanted to link to the Google Translate version, like I did in the past. Seems this is not possible anymore without a Google account:-(
For the more significant stuff -- where you go and when, who you communicate with and what you communicate about -- data from autonomous vehicles probably isn't going to be the main issue.
It will generally work a lot better -- cheaper, easier, more reliable, etc. -- if they get their surveillance data from a system designed for their purposes. And, of course, they don't have to wait for the widespread adoption of autonomous vehicles to build such a system. In fact, they haven't waited.
The big question is how our norms and laws around privacy reset after the explosion of information technology has wiped away all the assumptions on which our traditional notions rested.
Can you elaborate?
"Legally compelled to do so" means a warrant or a court order or something like that; which in my mind seems reasonable; in general, we do want courts to get evidence to determine what actually happened in (potentially fatal) accidents, instead of having to do without and have a greater chance of making a wrong judgement because they didn't have that information.
"If you outlaw freedom only outlaws will have freedom."
When you rent a safe deposit box the law doesn't get automatic access (warrant etc is required). They law could consider buying just a ride from an AV service to be "renting" the car as you would a safe deposit box, house, or apartment and require a warrant for police access.
But you'd have no authority over the cameras and other systems of all the other cars around you -- those would still be available.
Tricky.
I don't wonder -- it seems quite certain this will happen. It seems whenever new systems are put in place, surveillance is built in: Automatic bridge toll payment? Could have been done with a reloadable token but no, you have to tie it to a particular vehicle and identity, and that information has been used by law enforcement. Driver's licenses? They really need only a photo and an expiration date but instead they are belarded with all sorts of information (weight, address etc) which have nothing to do with driving.
In this case it will likely be shrouded by the overstretched and abused "third party doctrine" (if a third party has the info the govt can just go see it -- so the DEA doesn't have a license plate database, it can simply contract with a third party to make one instead). The government will simply require that AV fleets provide all their telemetry data including optical camera info.
> It seems to me that the infrastructure to support mass AV traffic would involve lots of sensors, cameras, and other electronic devices.
What sort of infrastructure should be built that is specific to AVs (and why should the government, rather than the fleet operators or other third parties implement it)? Why should they need even as much as humans require today?
How/why would a production car need to process it differently and why couldn't all processing be done in real time? Safety certification means the car's functionality has to be static, until the next certified software update. Data does not need to leave the car, since the purpose of the data is for the car to drive, which is a transient state.
The technology needs improvements, and the process of making those improvements is hungry for data. Every manufacturer will be in development/debug mode of their self-driving systems at least for the coming decade, so sensor data from every mile of every car is wanted and useful, and manufacturers are configuring production cars to upload as much raw sensor data as is practical.
Each society can decide whether to permit this particular product to collect unlimited data. The presumption is that the data belongs to no one. Let's see what happens when more than one large corporate entity lays competing claim to the same data subset, embedded by a wide-ranging data dragnet.
A 24/7 data feed from your toaster is not going to make your toaster better. It might help an R&D department identify some ways how the next toaster model should be different, but that's about it.
However, a self driving car is data-starved and is still going to be data-starved years from now. At any moment of time, your car could drive better and safer simply if it had more "experience" - the v1.24 software release can be meaningfully better than the v1.23 software release even if R&D department does nothing else but simply import the data received from millions of other cars; if your car is allowed to learn from what other cars saw.
Ensuring that the quality of driving systems increases as fast as it can is important for the society, with a major impact on injuries and casualties. I feel that it would be best for the society if we ensure that this learning (and the required data transfer) is not prohibited, as long as we can solve/restrict the potentially harmful uses of the data.
We don't actually know that self-driving cars are possible in unrestricted environments. The industry is asking for highly invasive surveillance but has no liability for failure to deliver. What happens if "more data" turns out to be insufficient? Will the next request be "restrict the environmental context" or "pass new laws to change human behavior"? Where does it end? Other industries have to deliver results before changing the market. Uber et al promised the moon to their investors, then ignored local laws, then promise to save lives, then ignore privacy concerns, then ..?