How connected cars might feed techo-feudalism

What exactly are we walking into and do we need a course correction?

Yes the headline is a bit polemic but bear with me, it’s thought provoking stuff. This post is another one that has been rolling around in my head awhile with a number of seemingly disparate threads coming together.

Late last year, Katharine Kemp, Associate Professor of Law at UNSW published an in-depth analysis of the data privacy terms for popular connected cars sold in Australia. The lengthy paper, Driving Blind - The Unexamined Privacy Risks of Connected Cars, was published alongside an accessible summary article, Modern cars are surveillance devices on wheels with major privacy risks. I read the article and the paper and realised I had just found another reason for not wanting to buy a new car any time soon!

Then just after Christmas, I was listening to the always thought provoking podcast The Good Fight and in the course of describing his ideas about the transition from capitalism to techno-feudalism (fascinating, go read the book), Yanis Varoufakis commented that Tesla “will soon make more money from the data collected by the cars than by selling spare parts for those cars.”

Not that I’d buy a Tesla (being a Starlink customer is compromising enough ever since Elon lurched from kinda cool to basically despicable ) but the idea that Musk is hoovering up data from every Tesla in the world at all times? Now that really did make me pause and want to revisit the connected cars phenomenon again.

What is a connected car anyway?

I’ve been driving a number of rental cars recently and I’ve noticed sometimes with pleasure, sometimes with irritation, that they tend to have a number of new ‘features’ that my somewhat venerable Toyota Prado does not. Love the ‘heads up’ holographic display of current speed. Not so much the lane following tech, that might work just fine in downtown Seoul but is downright dangerous in rural Tasmania.

My mother in law mentioned that her new car is very bossy and tells her to have a coffee break every 250km. And a friend related a funny but honestly-must-be-horribly-frustrating story about his sister needing to turn off all the safety indicators in her car every time she wants to back out of her own driveway because it’s a tighter turn than the car will ‘allow’ with all the sensors turned on (and of course they turn back on by default every time you restart the car).

But these joys/niggles aren’t really the connected car bit. Here’s the exec summary from Katharine’s paper

Connected cars are an emerging category of consumer vehicles marketed for features ranging from automatic SOS calls in the event of an accident and notifications when a child is left in the back seat, to the ability to cool a hot car in advance or take “selfies” of the driver whenever they smile. At the same time, connected cars act as highly sophisticated surveillance devices.

To provide their connected services, the vehicles automatically send data not just about the vehicle’s operation, but also about the driver and their family members, in real time to various overseas companies and Australian companies. This information can be sufficiently detailed and comprehensive that it creates a highly revealing picture of each driver, their real-time movements, risk profile, attitudes, values, moods, habits, family, associates, and activities.

Driving Blind: The Unexamined Privacy Risks of Connected Cars, Katharine Kemp

Typically, a connected car has a embedded SIM card and modem that transmits data via the mobile network. It’s turned on by default and somewhere between tricky and impossible to turn off.

The good stuff

There are definitely both safety and convenience features that come with a connected car. Emergency calls in the event of a bad crash could literally be a life saver. Stolen vehicle tracking could be a huge theft deterrent as well as a very useful tool when your car gets pinched.

Remote external and internal views of the vehicle would definitely be handy now and again. (Although take note that this also means any parking garage now potentially has tens to hundreds of remote surveillance devices at any given time. And there are scary implications for victims of domestic violence and coercive control.)

Note that I don’t mention here things like driver assistance with obstacle detection, etc. I’m willing to bet good money that that stuff doesn’t need real time connectivity and compute-on-device would actually be preferable in terms of both response time and reliability.

The not so good stuff

But. I’m guessing my readers have all lived long enough in this world to know that it isn’t going to just be the owners and drivers benefitting from this flow of data. This conflict has been brewing since the Mozilla Foundation published it’s own comparison of connected car privacy policies with a focus on the US market, in September 2023. I read that too, back in the day and remember feeling my toes curl.

More from Katharine.

In 2024, the Attorney General for the state of Texas brought proceedings against General Motors, making allegations concerning the “unlawful collection and sale of over 1.5 million Texans’ private driving data to insurance companies without their knowledge or consent”

This claim alleges that the company deceived many of its customers by compelling them to enrol in products such as “OnStar Smart Driver”, as part of its vehicle “onboarding” process and telling them that failing to enrol would result in the deactivation of their vehicle’s safety features, while concealing the terms of these data practices.

Earlier proceedings against General Motors included allegations that a consumer discovered that his driving behaviour data was being sent to the data broker, LexisNexis, after collection by General Motors, the manufacturer of his Chevrolet vehicle. The consumer claimed that LexisNexis then used this data to create a “risk score” which was cited by insurance companies as a reason that his car insurance was 21% more expensive than the previous year. According to the claim, the information transferred to LexisNexis contained data on 640 of the consumer’s trips, including start and end times, distance, speeding, hard braking and acceleration.

Use of some data some of the time is covered by Australian Privacy law. However if you live here and follow this space, you’ll know that there has been very slow progress on updating our privacy laws for the reality of the 21st century. And connected car data, like telco data, has a lot of grey areas. Is the location of the car at all times personal data? I’m sure you can find some very well paid legal teams to argue that it isn’t. But imagine what you could infer from it.

To summarise the main takeaway from the Driving Blind paper, it’s the same problem that society has had with data brokers for a long time, with connected cars tipping a whole lot of new data sources into the hopper: it is well nigh impossible to reliably de-identify granular data. We need to give consumers a lot more control over their data and truely informed consent.

Below is an excerpt of an eye opening table from the paper that gives a good idea of where this can all lead. I wish I could still believe that this was dystopian hand wringing but I think I stepped past that point with Cambridge Analytica.

If you can sell ads from it, someone somewhere is going to buy it.

The potentially quite scary stuff

Which takes us back to Yanis and his talk of techno-feudalism and Tesla making substantial amounts of money from the data collecting device that is the modern connected car. While I found both the podcast and his book very intriguing, I’m not enough of an economist to be able to write a decent summary of techno-feudalism so let me quote a recent article

Techno-feudalism is the idea that we are transitioning from capitalism to a system where tech companies function like modern feudal lords. Varoufakis argues that since the 2008 financial crisis, our economic system has fundamentally changed. The cloud, big data and digital platforms have become the “land” of this new era, controlled by tech giants like Google, Amazon and Meta.

These companies, Varoufakis contends, influence our behaviours and choices in ways that are not as voluntary as we might believe. Just as medieval lords controlled land and labour, tech companies wield immense power over data and access to digital spaces.

What is techno-feudalism?, The Beautiful Truth, October 2024

To quote Yanis directly “Jeff (Bezos, the owner of Amazon) doesn’t produce capital. He charges rent. Which isn’t capitalism, it’s feudalism. And us? We’re the serfs. “Cloud serfs”, so lacking in class consciousness that we don’t even realise that the tweeting and posting that we’re doing is actually building value in these companies.”

Hold that thought and hop on over to an article from 2020 about how Tesla harvests and builds on top of the data flowing out of the fleet of Teslas on roads around the world. Teslas owned by the person who purchased them of course, only maybe not owned quite as exclusive as we might have expected?

From the very beginning, Tesla has prioritized the collection and utilization of all possible data analytics from their car owners.

So, in essence, “Tesla owners are not just driving a car to commute to work or run errands. They are simultaneously training the Tesla AI/ML engines as they go,” according to George Paolini of CIO Magazine. As a result, Tesla has created “one of the most effective crowd-sourced AI/ML training initiatives around today.”

…

One of the other major factors fueling Tesla stock prices is their innovative approach to monetizing its 3 billion miles worth of driver data. For example, this past July, Tesla CEO, Elon Musk, announced it will be building a “major insurance company” that would use driver data to determine payment rates based on how aggressively or safely a person drove.

…

Another area of data monetization with high potential is hyper-targeted advertising. Tesla not only records human-car interactions and where a person travels to, but also records in-car conversations, texts and calls. Using this information, Tesla could easily structure and categorize this data to understand drivers' travel and consumption habits. Using these insights, they could display paid ads on the vehicles infotainment system or alert the driver to nearby places of interest via text, mobile app, etc.

Tesla: Automaker or Data Company?, AI Data & Analytics Network, Nov 2020

The implication that is hard to ignore from both Katharine’s paper and Yanis’ book is that driving a connected car, while it has some immediate and substantial advantages, has some quietly insidious, negative knock-on effects that most of us don’t even realise we’re signing up for.

Keep a Fergie in the shed

When you start looking, real world examples of possible futures aren’t hard to find. A couple of years back, Australia passed legislation requiring automotive manufacturers to share data required for service and repair of cars, part of what is commonly known as right to repair provisions. However this doesn’t cover agricultural equipment.

Breakdowns during harvest range from costly to devastating and a dealer callout could take hours to days. Not surprising then that most broadacre farmers have traditionally also been pretty competent mechanics.

However …

Agricultural machinery is not covered by right to repair legislation, which is why the modern farmer may have the latest John Deere working in the paddock and the old Fergie in the shed in case of breakdowns.

Most of the technology-reliant equipment that Australians use every day – from machines used in sectors like mining and health to household appliances and mobile phones – is not covered by right to repair laws. But additional issues for agriculture technology are the hefty price tags, the restrictions on on-farm maintenance and the distance to authorised repairers.

…

Wiseman says while copyright is a legitimate way to protect the software, the digital locks used by the manufacturers mean that it cannot be accessed for repair. “Once there is an operating system, you can’t just swap a part in and out because the system won’t accept it unless you have the diagnostic software,” she says.

This leaves farmers such as Honner frustrated. Breakdowns, most likely at the critical point of harvest, mean machinery can be out of action for an indeterminable amount of time and tinkering is not encouraged. “With new machinery there is pressure on us not to do our own maintenance – even an oil change, and that’s not brain surgery,” Honner says. “I feel we’ve been dumbed-down.”

‘We’ve been dumbed-down’: Australian farmers want the right to repair their own tractors again, The Guardian, Jan 2025

So what next?

Cultivate being an aware consumer. That’s what it boils down to. Buying a new car? Choosing a new phone? Fitness tracker? Vacuum cleaner? Doorbell? Opt in mindfully and educate others where you can. Lobby your policy makers if that’s your thing.

Data use / misuse isn’t a given, there are plenty of choices and alternatives and we need transparency and balanced debate.