Cars, Cameras and Data collection
By: Evan Judge
I like cars. I learned about the mechanics of cars, how to weld, how to change out a camshaft, when I was in college from a family friend. I enjoy the automotive industry. I love how Kia was the first car company to drop cigarette sockets for USB plugs, and the story behind why old French cars have yellow headlights. There was a time when innovative technology in cars meant retractable seat belts or pop up headlights. Now automotive technology is moving at 88 mph to take us further with more convenience, but the pace of development causes me at times to stop and think about the repercussions. Cars are changing from mechanical systems to cutting edge tech devices. Just look at all the self driving car startups.
Earlier this month, Tesla announced that they would be requesting to “collect snippets” of video feeds from the 8 cameras on the cars to better analyze recognition of lanes, signs, and streets. Tesla says that the information won’t be connected to the car’s VIN (Vehicle Identification Number), and that no PII (Personally Identifiable Information) will be passed along as they analyze and share the data with third parties. However, there is still room for intrusions of privacy. Tesla doesn’t specify how long the clips are, how often they are taken, or how disaggregated the clips are from the other cameras on the car. [Note: I reached out to Tesla with questions about the video being captured. At the time of writing, I hadn’t heard back.] There is still potential for drivers to be identified, whether by routes recorded during transit or destinations captured on camera (i.e., work, home), potentially negating efforts to protect privacy.
This development reminded me of some challenges in the data privacy world that illustrate possible issues:
1) What a lovely assemblage of small pieces of your life!
The “Mosaic Theory” of the Fourth Amendment was first mentioned in United States v. Maynard (the D.C. Circuit case leading to the Supreme Court case US v. Jones on GPS tracking), where the court found that “individual investigatory steps taken by law enforcement do not amount to a Fourth Amendment violation, but when viewed in the aggregate…infringe on a person’s reasonable expectations of privacy.” Basically, by taking snapshots of location data over a period of time, law enforcement was able to put together a picture of the defendant and their life. No single snapshot revealed anything specific or particular, but when viewed as a whole a clear picture was presented. Much like a Mosaic.
While the Tesla development is not a 4th amendment issue, it still follows the same pattern of individual pieces of data forming a mosaic to give a picture of identification. The snippets could be used to ID a person based on driving routes and habits, and what the cameras capture. NYU had a great article over Mosaic, Machine learning and tracking in 2014.
2) I always feel like, ♫ somebody’s watching me ♫
Cameras are everywhere, and the world pretty much accepts that everything is always being recorded. But in certain circumstances, a party still needs permission from others before they roll tape. In 2014, GM released a Corvette that captured interior video when it was in “Valet Mode.” Valet mode restricts the 400+ HP sports car from accelerating quickly, puts the radio to a low volume, and limits the car from going over 45 miles per hour. It was hailed as a baby monitor for your car. The idea is to prevent valets from re-enacting Ferris Bueller’s Day Off. To further counteract this, the Corvette would record the whole time the valet key was used from the exterior and interior camera. However, this was an issue in the 12 states that require two-party consent for recordings. The car owner often would not inform a valet of the recording before handing over the keys, so GM had to warn owners it may be illegal and consider a recall of the model.
“The problem relates to laws in 12 states, including California and Michigan, that ban audio recording of private conversations unless all involved parties provide their consent. In some of the states, people who make unauthorized recordings can be charged with a felony.”
It appears that car companies can often move too fast with new tech that makes sense in use, but do not follow the standards for compliance. Yes, Tesla is vastly different than GM, but still, I am wondering if there has been a proper check for legal complications from the recording and passing along of theses video snippets.
3) “Alexa, who dunnit?”
Alexa is one of my favorite tech tools. She is always listening, and ready at a few syllables beckoning to answer questions. But while this trove of recording data that Amazon has amassed may seem like an asset, it can also be a liability. In 2017, Amazon had their Alexa recording data subpoenaed because the device may have overheard a murder. The police wanted to review the possible recordings since there was an Echo in the room where the murder happened, and Amazon fought the request.
In February – “Arkansas police recently demanded that Amazon turn over information collected from a murder suspect’s Echo. Amazon’s attorneys contend that the First Amendment’s free speech protection applies to information gathered and sent by the device; as a result, Amazon argues, the police should jump through several legal hoops before the company is required to release your data.”
Eventually, the suspect ok’d the release, and Amazon complied, yet again leaving the legal world without a precedent in a technology-related case. So what happens when a Tesla is in view of a car wreck or gas station robbery? Will the videos be subpoenaed?
Tesla’s new video snippet collection is an interesting idea that could certainly help improve autonomous offerings, but it presents some fascinating privacy concerns. America gets plenty of things wrong when it comes to automotive tech. Why do I still need side view mirrors when it’s more efficient to use cameras? Why can’t I have laser headlights? We sometimes take a little too long to bring the best to our daily commute.
What I hope we do get right is the consideration of what is best for consumers from a protection and feature standpoint, while still being cognizant of the potential negative outcomes.