That, of course, raises all kinds of legal concerns. In the article linked above, author David Kravets rightly points out the value of that data in criminal and civil cases, as well as the more remote possibility of mining the data to issue traffic citations. When the car's owner or driver is the defendant, that would certainly seem to challenge "The right of the people to be secure in their persons, houses, papers, and effects."
But this isn't really a technology problem. If that data was stored on something like a flight data recorder in your car—and it in fact is in most newer cars—the Fourth Amendment uncontroversially protects it. Probable cause, in most cases supported by a warrant, would be necessary to use the data in a criminal case. The reason is simple: you own that data.
Here's where OnStar's policy becomes a problem: they own the data. Like most software and services, the driver of the car does not "own" the OnStar system; the software and right to use the service is licensed from OnStar and the service is bound by its terms of service.
That's pretty typical of the contemporary economy. As we've moved from a commodity economy to an intellectual economy (i.e., one in which the limitation on goods is the intellectual act of creating the good itself and not the reproduction of it, as in software or music), we've also moved from an ownership-based economy to a licensing-based one. Many of the things we use in our daily lives do not belong to us. We merely have the right to use them.
That's a problem for traditional search and seizure protections. Those protections are rooted in the idea of ownership: the "persons, houses, papers, and effects" owned by a (potential) defendant could not be searched seized without probable cause, because that was really all there was to a defendant. The problem of what third parties could be compelled to provide about that defendant, constituted a relatively minor area of exception, often understood as unproblematic by noting a consensual transaction between the defendant and the third party. Ownership is the basic tool for protecting the individual in classical liberalism.
In the licensing economy third party data is extensive, and consent is made exceptionally problematic by the ability of licensors to change terms essentially at will, our practical dependence on such services (how would your boss
take it if you decided you weren't going to have Internet service at
home?), and the near complete inability to avoid onerous license terms for such services (try finding a cell phone license that doesn't allow the provider to collect and store your location data). It is legally true but practically false that I have freely consented to the terms of service for many of my possessions; if I refuse to consent to them I would find it exceptionally difficult to participate in contemporary American life.
So as we move to an intellectual economy, it becomes necessary to rethink whether the Fourth Amendment still protects my person and property from unreasonable intrusion by the state. A key part of the political superstructure of liberal democratic societies breaks down when the basic structure of possession is licensing rather than ownership; an ownership-based Fourth Amendment cannot protect "The right of the people to be secure in their persons, houses, papers, and effects" because many of the threats to these things are no longer under the people's control. My security rests in the hands of people who have no incentive to protect it, and much incentive to cooperate with the state.
Solutions? You're guess is as good as mine here. I think there's a larger project to be pursued in broadly rethinking how our the ideas, structures, and practices of an industrial political economy fit into an intellectual economy. Goods produced with no variable costs break down the pricing model under the norms of a perfectly competitive economy (you know, the ones that justify the assumption that markets are efficient rather than rent-seeking). Licensing brings about extensive conflicts between producers and consumers over what "property" means. The organization of open-source development communities looks nothing like traditional business enterprise, is composed of people bound by intellectual rather than economic interests through personal involvement rather than contract, and involves a very different understanding of freedom.
I think all of these things represent cases in which the ways in which goods are produced initiates a transformation in the organization of the economy, polity, and society. Those aren't deterministic; I see these forces as constraints on rather than determinants of structures. But a structure that is incompatible with how the goods are produced will be at least a source of conflict, one in which I suspect the economy is heavily favored.