No Right-to-Repair for Your Self-Driving Car
Ownership matters, but safety matters more
A final reminder: Changing Lanes will begin reserving most of its content for paying subscribers one week from today. Pledge to pay now, at this link, and enjoy Charter Membership rates of $8 per month or $80 per year; that’s a 20% discount on regular rates, which will begin on 6 May 2025.
It's 2035. You've owned a Studebaker Grizzly luxury car for two years. On your last trip, you noticed the lane-keeping system has stopped working, causing the car to drift over the line. The closest Studebaker service center is 250 km away with a three-week wait. Your LLM assistant, Mercury, contacts a local shop that could fix it tomorrow, but reports back with disappointing news: Studebaker has an algorithmic lock on your car's software. While the lock is on, the shop can't fix it, and if they break the lock, Studebaker will remotely disable your car, claiming any unauthorized modification makes it unsafe.
As ADAS, and ultimately wholly automated-driving systems (ADS) become commonplace, stories like this will cease to be fiction. It’s already the case for other products that manufacturers insist on retaining absolute control of their systems, to ensure they operate safely. And users already insist on their right to repair their own devices, not only on principle, but also because abandoning it allows manufacturers a free hand to exploit their customers.
Both sides have a good point, but only one side can be satisfied. What should we do?
I was interested to read a recent article on the subject: The Right to Machines' Souls by
in Arena magazine (which also published my own piece, Taking Our Hands Off the Wheel, in an earlier issue). Rob argues that the right to repair one’s own devices represents a fundamental American value, and one central to consumer autonomy and well-being. Rob is a friend of Changing Lanes and, like me, a Fellow of Institute; but both Rob and the RPI welcome intellectual debate.That means that I can freely express my view, which is that at least as applied to driving automation, Rob is incorrect.
Driving automation presents unique safety challenges that require special consideration, and as such, there should be no consumer right-to-repair for ADS or ADAS.
But before I make that case, let’s review the argument that Rob makes.
The Right to Machines' Souls
The formal right-to-repair movement has its roots in amendments made in 1990 to the USA’s Clean Air Act. The original 1970 legislation aimed to reduce harmful emissions into the atmosphere, and to that end required auto manufacturers to include exhaust-monitoring technology in their vehicles. They recognized thereafter that this gave manufacturers an opportunity to abuse their customers by creating monopolies on repair of that tech. Accordingly, the 1990 amendment to the legislation required automakers to share diagnostic tools and information relevant to these computers with independent repair shops. California went further, mandating that this information be provided “at cost and terms that are no greater than fair market value”.
This legislation only covered emissions-related components, but it was a precedent, and one that attracted attention as other parts of modern cars become software-dependent. By 2001, independent shops routinely had to turn customers away because they lacked the diagnostic tools for modern vehicles. The American Service Association estimated that $18 billion in sales to independent shops were being lost annually due to proprietary telematics, which was roughly 10% of the auto repair market at that time.
The situation reached a turning point in 2012 when Massachusetts introduced both legislation and a ballot initiative on right-to-repair for motor vehicles. The ballot measure passed with an overwhelming 86% of the vote, illustrating widespread public support. This victory was decisive; with Massachusetts having taken this ground, there was no way to halt the dissemination of the relevant tools and knowledge nationwide. Recognizing this reality—that ‘information wants to be free’—within two years, a coalition of automotive manufacturers agreed to make diagnostic and repair information available, to owners and independent repair shops, on “fair and reasonable terms”.
Rob frames the right-to-repair as an expression of self-sufficiency, a core American value. When consumers purchase a product, they should have complete rights to understand how it works, modify it according to their needs, and fix it when it breaks, or when the manufacturer declares it obsolete. Rob calls these rights the soul of ownership.
Rob offers several aggravating examples of how, in recent years, this kind of ownership hasn’t been respected. He points to American farmers struggling with John Deere equipment restrictions, and quotes one farmer: “You want to replace a transmission and you take it to an independent mechanic—he can put in the new transmission but the tractor can't drive out of the shop. Deere charges $230, plus $130 an hour for a technician to drive out and plug a connector into their USB port to authorize the part”.
This is inconvenient, certainly. It may also be exploitative: is Deere overcharging? How would one know? Certainly it seems likely that the firm is; given their monopoly on authorization, they aren’t subject to market discipline.
Similarly, McDonald's franchisees find themselves unable to repair their own McFlurry machines because the parent corporation has a “copyright” on the technology. Only a McDonald’s-authorized technician is permitted to do the work; but the cost of such work is greater than the revenue that McFlurrys bring in, so many restaurants don’t bother to do so, meaning consumers deprived of milkshakes and franchisees of the revenue they could have earned.
Rob doesn’t mention it, but his stories remind me of BMW's attempt to introduce a heated-seats-as-a-service subscription. In 2022, BMW began charging customers a monthly subscription fee to activate heated seats that were already physically installed in their vehicles. This wasn't about safety or functionality; it was purely about extracting ongoing revenue from hardware the customer had already purchased. Customer backlash was severe, with many correctly perceiving this as double-charging, and as legislators started to decry the practice and threaten a response, in 2023 BMW abandoned the practice.
Rob sees cases like these as examples of service monopolies, artificial barriers to repair that extract ongoing revenue from products already purchased. It’s hard to argue with him. Certainly many politicians see it his way: following Massachusetts’ lead, right-to-repair legislation has been passed in California, Colorado, Oregon, and Minnesota. In Canada, the federal government passed nation-wide right-to-repair laws in 2024. Consequently, in Canada, though not yet the USA, the McFlurry flows freely.
Rob sees the contest over the soul of ownership as a battle between two American archetypes, namely the ‘rugged individual consumer’ and the ‘iconoclastic creator’. Both embody the best of American individualism, and ideally both could be accommodated, but their interests fundamentally diverge when it comes to repair rights. Rugged individual consumers desire self-sufficiency and complete control over their purchases, while iconoclastic creators seek the freedom to design products that maximize performance and minimize constraint, which may require making repairability the lowest priority.
The iconoclastic creator who casts the longest shadow is certainly Steve Jobs. Famously (notoriously?), he designed Apple products, from the iPhone all the way back to the original Mac, to be seamless integrations of hardware and software, but ones that consumers couldn’t even open, let alone fix. Consumers were welcome to play in Apple’s walled gardens, but not to plant their own flowers.
Notwithstanding the spread of right-to-repair laws, iconoclastic creators—and following behind them, grasping monopolists—enjoy the balance of power today. Through software, manufacturers can now monitor and control products remotely long after sale. “Part pairing” techniques ensure only official components work with devices, and over-the-air updates can block third-party repairs or modifications, even to products the manufacturer refuses to support. In one egregious example Rob offers, D-Link refused to patch 60,000 routers with a critical security flaw because the company had declared them at “end-of-life”, effectively forcing consumers to upgrade despite the devices being otherwise functional.
Rob's central claim is that in our connected world, the soul of a machine is shared among the creators and the owners, but the balance has tilted too far toward manufacturers, undermining consumer autonomy and choice. While he acknowledges the complexity of competing interests, he ultimately argues for stronger consumer-repair rights. That was where the balance was struck in an earlier day, when machines were mechanical, and it should remain there today when machines are electronic and, increasingly, digital. Firms’ efforts to restrict how consumers use their purchases would be bad enough were it just a cash grab, but it’s worse than that; it’s also an intolerable denial of freedom.
So far, I've been nodding along with Rob's argument, agreeing with his defense of consumer rights against corporate overreach. Perhaps you have as well. But as I consider ADAS and ADS, I find myself obliged to take the manufacturers’ side. For most kinds of good, I think Rob has the right of it. But I also think that automated driving systems are a special case. In this domain, denying owners the freedom to modify or repair them is the right move.
The Right to Not Die from Another’s Negligence
To illustrate why that’s so, let’s briefly review what makes up an ADS, or its poor cousin, an ADAS.
Unlike conventional vehicle components, these systems integrate a complex array of sensors, sophisticated software, and specialized mechanical elements that work together as a unified system. A typical ADAS relies on multiple sensor types: cameras that “see” the road, combined with radar that detects objects even in poor visibility, and ultrasonic sensors for close-range detection. Moreover the true ADS we’ve seen to date rely on lidar as well; you can’t buy a vehicle equipped with true ADS today, but surely someday there will be private vehicles running on Waymo or Zoox software, meaning that lidar will be part of the package.1 These sensors generate enormous amounts of data that must be processed by specialized computing hardware running advanced algorithms.
These components aren't merely assembled; they're calibrated together as a system. A camera must be precisely aligned to match its field of view with the vehicle's understanding of its position. Software must know exactly how to interpret signals from each sensor. Actuators that control steering, acceleration, and braking must respond precisely to software commands.
It’s this integration that makes ADAS and ADS, and their constituent parts, different from traditional vehicle components. At first glance, it may seem natural that if a vehicle sensor breaks, the owner or a third party could replace it; a cracked forward-facing camera or a faulty radar emitter could be swapped out in the same fashion as a door handle or a transmission. But at second glance, we see the critical challenges of calibrating and integrating those components.
What if the replacement camera has slightly different specifications, or the radar unit is mounted at a slightly different angle? These changes could fundamentally alter the ADAS’ perception capabilities. What if a repair is botched? What if an aftermarket part fails? The failure might not be obvious; ADAS errors can be subtle and situational. A misaligned sensor might perform adequately under ideal conditions, with the software compensating for small discrepancies, but fail in a particular edge case. And that failure might emerge weeks (or thousands of miles) after the responsible modification, making causation difficult to establish.
Worse, the consequences of such a failure would likely be felt by other road users. When a conventional vehicle breaks down due to poor maintenance, it typically stops moving, which is risky but relatively manageable; it’s not a fail safe, but at least a fail-more-safely. Conversely, an ADAS error is neither; it can mean the vehicle making the wrong moves at full speed.
The problem extends beyond sensors. In modern Mercedes vehicles, the windshield is an integral part of the ADAS system because cameras and sensors are mounted to it and ‘see’ through it. Replacing that windshield with an aftermarket part that has slightly different optical properties or mounting points could compromise the entire system's functionality in ways not immediately apparent, such that Mercedes forbids any modification to its proprietary glass.
Traditional right-to-repair principles presume that the risks of improper repair primarily affect the owner who chooses to accept those risks. With ADS, the consequences extend far beyond the individual, creating a compelling public interest in ensuring that all modifications and repairs maintain rigorous safety standards, even if that means placing limitations on conventional repair rights.
Threading the Needle
How can this be done? It’s instructive to look at how this issue is handled in parallel fields. In aviation, for example, there was the 1989 crash of Partnair Flight 394. In that case, third-party bolts, improperly heat-treated during manufacturing and as such only half as strong as specified, failed in mid-flight; as a consequence, the tail fin snapped off, causing a crash that killed everyone on board. In healthcare, the FDA has documented a case where the third-party parts used to repair a colonoscope also snapped off, exposing sharp edges that injured the patient.
These industries have responded with rigorous regulatory frameworks. The aviation industry still allows third-party companies to make replacement aircraft parts, but only after undergoing stringent design and production approval processes ensuring they meet original specifications and airworthiness standards. Aircraft mechanics and repair stations must be FAA-certified. Similarly, in healthcare, medical device servicing is tightly regulated, with the FDA requiring proper training, documentation, and quality systems for anyone servicing medical equipment. To ensure patient safety, third-party servicers must typically meet the same ISO certification standards as the original manufacturers.
In other words, parallel fields neither allow manufacturers to monopolize repairs, nor allow complete freedom to modify safety-critical equipment. Instead, they establish and maintain certification frameworks that permit third parties to service components, but only those who have demonstrated and maintain rigorous safety standards.
Something similar will be required for ADS and ADAS work.
These standards for aviation and medical devices provide an instructive model for ADS and ADAS repair. What's needed is a certification framework that permits independent repair while ensuring safety standards are maintained. Such a framework might include:
Firstly, a tiered certification system for repair personnel. Just as pilots, electricians, and doctors have licensing levels, ADAS maintenance could be stratified. Core sensor calibration or software updates would require certification; importantly, this means that owners have no right-to-repair of their own (absent certification).
Secondly, for vehicles sold to the public, regulation must require standardized interfaces and diagnostic tools that provide transparency while protecting core safety functions. Much like the OBD-II port standardized access to emissions systems in the 1990s, a standardized ‘ADAS diagnostic port’ could allow third-party tools to safely interact with ADAS functions.
Thirdly, a rigorous validation process following repairs. After any critical repair, the ADAS should enter a ‘safe mode’ requiring a suite of self-tests before resuming full operation, and the results documented and stored somewhere accessible to manufacturers and regulators.
Returning to our opening scenario, the solution isn't giving Studebaker (or any ADAS vehicle manufacturer) carte blanche to lock out all repairs, nor is it allowing anyone, even the car’s owner, to tinker with safety-critical systems. The better outcome would be if that local shop’s technicians were certified in maintaining Studebaker's ADAS, with standardized tools to diagnose and repair the lane-keeping issue safely. You, the owner, could get your car fixed conveniently and affordably; Studebaker maintains confidence that repairs meet safety standards; and everyone on the road benefits from confidence that your ADAS will function properly.
This approach respects the legitimate safety concerns of manufacturers while preventing them from abusing that power to create service monopolies. It acknowledges that the soul of a machine is shared, not only between creator and owner, but in this case with the public. In a world where cars will increasingly drive themselves, the right to repair will balance against the right of everyone else to travel safely beside them.
___________
Respect to
and for comments on earlier drafts; please note that Rob’s review does not imply his agreement! Please also note that this piece depends in part on earlier research and writing that I undertook with , in a proprietary report that, sadly, is no longer freely available.Just in the past few days, Alphabet CEO Sundar Pichai mentioned, on Alphabet’s most recently earnings call, that as regards Waymo, “There's future optionality around personal ownership as well.” So this scenario may indeed come to pass.