Published at 70 Stanford Law Review 99 (January 2018) by Alan Z. Rozenshtein, then visiting assistant professor at the University of Minnesota Law School and, at the time of writing, an attorney advisor in the Office of Law and Policy at the Department of Justice's National Security Division. The article carries a disclaimer that it reflects the author's views alone and was reviewed by DOJ to prevent disclosure of classified information.
The gap it identifies
The article's premise is that the surveillance-regulation literature had been analysing the wrong set of actors: "scholarly attempts to answer this question have suffered from a serious omission. Scholars have ignored how government surveillance is checked by surveillance intermediaries: companies like Apple, Google, and Facebook that dominate digital communications and data storage and on whose cooperation government surveillance relies."
The framing case is Apple's 2016 dispute with the FBI over the San Bernardino iPhone. The article presents itself as "the first comprehensive analysis of how surveillance intermediaries constrain the surveillance executive: the law enforcement and foreign-intelligence agencies that conduct surveillance."
The structural insight is that surveillance depends on cooperation from private parties, which makes those parties a check whether or not any law designates them as one.
Three techniques of resistance
Intermediaries "have financial and ideological incentives to resist government requests for user data," and exercise it three ways:
- Proceduralism and litigiousness — approaches that "reject voluntary cooperation in favor of minimal compliance and aggressive litigation." The resistance operates through the letter of legal process rather than around it.
- Technological unilateralism — "companies design products and services to make surveillance harder." This is the only technique that does not run through any institution, and it is the one the article's normative argument turns on.
- Policy mobilization — "rall[ying] legislative and public opinion against government surveillance."
The surveillance separation of powers
Intermediaries also work indirectly, by activating other constitutional checks: they "make the surveillance executive more subject to interbranch constraints from Congress and the courts and to intrabranch constraints from economic and foreign relations agencies as well as from the surveillance executive's own surveillance-limiting components."
The intrabranch point is the less obvious half — resistance from firms strengthens the hand of agencies within the executive whose mandates cut against surveillance expansion, including economic and foreign-relations agencies and internal compliance components.
The normative argument
The article's conclusions are deliberately two-sided, which is what distinguishes it from advocacy on either side.
It introduces the surveillance frontier — "the set of tradeoffs between public safety, privacy, and economic growth from which we choose surveillance policy" — and holds that intermediaries "can both improve and worsen" it.
On surveillance self-government, meaning "the democratic supervision over surveillance policy," the split is between which technique is used: intermediaries "enhance surveillance self-government… when they mobilize public opinion and strengthen the surveillance separation of powers," but "undermine it when their unilateral technological changes prevent the government from exercising its lawful surveillance authorities."
That distinction — between resistance that routes through democratic institutions and resistance that forecloses their output by technical means — is the article's central contribution, and it transfers directly to questions about whether AI developers' unilateral design choices about model capabilities and refusals are a check on state power or a substitute for democratic decision.
Relationships
- supports: Alan Z. Rozenshtein — author of the work
- related: AI and Surveillance — supplies the intermediary framework the AI-era analogues are read against
- related: Surveillance Technology, AI and Privacy, AI Power Concentration, ICE — Immigration and Customs Enforcement (AI Deployer)