Great piece. As you write more in this series, how should the failure to reform American liability law for human beings shape what liability we adopt for AI? For example, it would be strange to hold an AI self-driving car (and its manufacturers) to a different standard than one driven by a human being. One could make similar arguments for copyright standards and other areas of law. These all desperately need reform, but does that mean we should adopt a better policy regime for AI before we've reformed the policy regime for humans?
In short: you've convinced me that our liability regime needs an overhaul. If it is not overhauled, though, wouldn't it be unjust to argue the AI world ought to be held to different standards?
My hope is that AI can serve as a proving ground for new types of policy regimes in general, and that those can gradually be expanded to non-AI areas of life (or just subsume the old systems as AI becomes a larger part of life).
You deftly illustrate why I can’t take seriously lawyers’ hoary admonitions that jurors are “finders of fact”. That may nominally describe their function but their findings frequently seem to be devoid of factual analysis.
Even if they do "find fact," their ability to analyze those facts and make conclusions about complex technical matters is questionable--that was never really the point of trial courts and they have performed poorly at the task.
a) "It is a system that is, as far as I know, unique to the United States."
Not sure what you mean here—tort liability definitely isn’t unique to the US. Are you talking about the scale of liability, the way it’s applied, or something else?
b) "A key function of courts is to enforce contracts."
Is that really the main function of courts? Courts don’t just blindly enforce contracts—they mostly try to assess the dynamics between the parties, including bargaining power. They step in when something has gone wrong, not just to rubber-stamp agreements.
c) "It can obtain regardless of whether the harm was 'caused' by a manufacturing defect or by a design defect."
That doesn’t seem quite right—strict liability still requires causation. You can’t just say, “this product exists, and I was harmed,” without proving some causal link.
d) "It can obtain, in many cases, regardless of whether a regulator said the product was safe."
Regulators make mistakes—that’s not outrageous, that’s reality. The italicisation here makes it sound like strict liability is unfair when it contradicts regulators, but isn’t that the point? If a regulator calls something safe and it still causes harm, why shouldn’t courts step in?
e) "Strict liability attacked at its core the way that transacting parties had apportioned risk and responsibility among themselves for centuries."
Maybe that’s a good thing? Power imbalances in contracts are real, and the average person often has no real choice or understanding of the risks they’re taking on. Courts stepping in to correct that seems like a feature, not a bug.
f) "Strict liability initiated—begged for, really—a crusade against large American companies, regardless of how much harm they created."
Wait—how does this follow? If someone crashes into my car, I can’t sue the car company unless their design actively contributed to the harm. There still has to be causation. Strict liability doesn’t just mean “sue the richest company involved.”
g) "The liability revolution was a failure."
That’s a huge claim, and it feels like it’s missing the flipside of the argument. Sure, there were unintended consequences, but didn’t it also make products safer and companies more accountable? Feels like a pretty one-sided take.
h) "Reasonable care" standard.
The problem is that AI labs will always be able to justify that they took "reasonable care." Courts don’t have the expertise to evaluate whether a lab's safety measures are actually meaningful. If this is the standard, it won’t change anything—it’s just the status quo dressed up.
Lots of questions here so feel free to respond to whichever, I expect to have made some errors in understanding what you wrote so keen to hear those! Take care.
1. Strict product liability is unique to the United states.
2. Contract enforcement is “a key” function of courts, not the only one.
3. What I meant was “liability can obtain whether the defect was in manufacturing or design”
4. That’s what tort liability activists would say too, but the question you need to seriously consider is: how suited are judges and juries to make increasingly technical decisions about this stuff? The empirical answer is not very.
5. Maybe! As I say, I am sympathetic to the problems with old school contract.
6. You absolutely can sue the car company because “design contributing to crash” is extremely nebulous and hard to adjudicate. There are hundreds of thousands if not millions of cases like this.
7. I cannot prove a negative (ironically, one of the many problems with tort liability litigation), so I can’t really answer this question beyond what I wrote.
8. Yep! Liability isn’t supposed to be a policy instrument to affect radical change in corporate life in my view.
Not being familiar with that law I cannot say, but I would be surprised if product liability is as broad as ours (especially ours at the peak of the liability revolution). It is true that anglo countries that have inherited common law tend to have strict liability in more places.
Impressive work!
Great piece. As you write more in this series, how should the failure to reform American liability law for human beings shape what liability we adopt for AI? For example, it would be strange to hold an AI self-driving car (and its manufacturers) to a different standard than one driven by a human being. One could make similar arguments for copyright standards and other areas of law. These all desperately need reform, but does that mean we should adopt a better policy regime for AI before we've reformed the policy regime for humans?
In short: you've convinced me that our liability regime needs an overhaul. If it is not overhauled, though, wouldn't it be unjust to argue the AI world ought to be held to different standards?
My hope is that AI can serve as a proving ground for new types of policy regimes in general, and that those can gradually be expanded to non-AI areas of life (or just subsume the old systems as AI becomes a larger part of life).
You deftly illustrate why I can’t take seriously lawyers’ hoary admonitions that jurors are “finders of fact”. That may nominally describe their function but their findings frequently seem to be devoid of factual analysis.
Even if they do "find fact," their ability to analyze those facts and make conclusions about complex technical matters is questionable--that was never really the point of trial courts and they have performed poorly at the task.
Hey, few notes/confusions:
a) "It is a system that is, as far as I know, unique to the United States."
Not sure what you mean here—tort liability definitely isn’t unique to the US. Are you talking about the scale of liability, the way it’s applied, or something else?
b) "A key function of courts is to enforce contracts."
Is that really the main function of courts? Courts don’t just blindly enforce contracts—they mostly try to assess the dynamics between the parties, including bargaining power. They step in when something has gone wrong, not just to rubber-stamp agreements.
c) "It can obtain regardless of whether the harm was 'caused' by a manufacturing defect or by a design defect."
That doesn’t seem quite right—strict liability still requires causation. You can’t just say, “this product exists, and I was harmed,” without proving some causal link.
d) "It can obtain, in many cases, regardless of whether a regulator said the product was safe."
Regulators make mistakes—that’s not outrageous, that’s reality. The italicisation here makes it sound like strict liability is unfair when it contradicts regulators, but isn’t that the point? If a regulator calls something safe and it still causes harm, why shouldn’t courts step in?
e) "Strict liability attacked at its core the way that transacting parties had apportioned risk and responsibility among themselves for centuries."
Maybe that’s a good thing? Power imbalances in contracts are real, and the average person often has no real choice or understanding of the risks they’re taking on. Courts stepping in to correct that seems like a feature, not a bug.
f) "Strict liability initiated—begged for, really—a crusade against large American companies, regardless of how much harm they created."
Wait—how does this follow? If someone crashes into my car, I can’t sue the car company unless their design actively contributed to the harm. There still has to be causation. Strict liability doesn’t just mean “sue the richest company involved.”
g) "The liability revolution was a failure."
That’s a huge claim, and it feels like it’s missing the flipside of the argument. Sure, there were unintended consequences, but didn’t it also make products safer and companies more accountable? Feels like a pretty one-sided take.
h) "Reasonable care" standard.
The problem is that AI labs will always be able to justify that they took "reasonable care." Courts don’t have the expertise to evaluate whether a lab's safety measures are actually meaningful. If this is the standard, it won’t change anything—it’s just the status quo dressed up.
Lots of questions here so feel free to respond to whichever, I expect to have made some errors in understanding what you wrote so keen to hear those! Take care.
Good questions!
1. Strict product liability is unique to the United states.
2. Contract enforcement is “a key” function of courts, not the only one.
3. What I meant was “liability can obtain whether the defect was in manufacturing or design”
4. That’s what tort liability activists would say too, but the question you need to seriously consider is: how suited are judges and juries to make increasingly technical decisions about this stuff? The empirical answer is not very.
5. Maybe! As I say, I am sympathetic to the problems with old school contract.
6. You absolutely can sue the car company because “design contributing to crash” is extremely nebulous and hard to adjudicate. There are hundreds of thousands if not millions of cases like this.
7. I cannot prove a negative (ironically, one of the many problems with tort liability litigation), so I can’t really answer this question beyond what I wrote.
8. Yep! Liability isn’t supposed to be a policy instrument to affect radical change in corporate life in my view.
1. I don't think that's true? The Consumer Protection Act 1987 for example does have strict product liability in the UK for eg.
No strong comments on the rest of what you said. Thanks for your response.
Not being familiar with that law I cannot say, but I would be surprised if product liability is as broad as ours (especially ours at the peak of the liability revolution). It is true that anglo countries that have inherited common law tend to have strict liability in more places.