Edition 013: The Taking of Human Data
The law has names for some of what a person produces by living. We have referenced some of these names before—personal data, biometric identifiers, protected health information, genetic information, intellectual property, anonymized data, and aggregated data. Each name comes with a definition, and each definition generally comes with a set of rules about what the holder of the data can and cannot do with it. These categories are too small. They were drawn around pieces of a life by people who were not looking at the whole life and the whole person.
So I am going to use a different term. Human data is what a person makes by being born and living a human life. Every search, every purchase, every message, every genetic test result, every fingerprint, every heart rate, every location, and every keystroke. The record traces the life. It is not separable from the life that produced it.
Academic writers have spent decades analyzing pieces of this.[1] The point I want to add is a systemic one about how the law has failed us and favored the corporations. Law is a human institution. Humans built it, and humans decide, generation by generation, what it is for. For most of its history, American law was built to serve a narrow set of humans, the ones with property, and to exclude the rest. The corporation is a tool that law made for an even narrower set, the holders of capital, so that capital could be pooled and risk could be shared. The tool is doing what it was built to do. It has also grown into something the law’s drafters did not foresee. The largest corporations now have more capital, more reach, and more capacity to shape information than most governments. The law that made them is still the law that governs them, and the law is slower than they are.
The argument I want to make is that the rest of us are inside the law too, and that the law owes us something. The human data the corporations have taken is still ours. I will spend the rest of this edition explaining why. They have it because there is no law stopping them or requiring them to consider anyone but themselves. And the corporations are not waiting for the law to catch up. They are writing the law, through lobbyists and regulators, and the terms of service that the rest of us click through. A law that lets a tool built for some take from all the others, treats the taking as legitimate, and is increasingly written by the tool itself, is a law that has decided who counts and who does not. That is a choice. It can be made differently.
***
The corporation has a life too, of a different kind. As discussed in Edition 010, section II, the corporation is a person the law invented so that capital could be pooled and risk could be shared. It is an artificial person under the law, and what it produces is what I want to call artificial data.
Its financial statements, its forecasts, and its compliance reports are records of its invented existence. They are not the record of a life. They are the record of an entity the law made up. The legal categories that protect data as a corporation’s property are also categories the law made up for the entity it made up. The whole construct is internally consistent. It is not real. An artificial person produces artificial data about its artificial life, and takes human data as its own, and the law treats all of it as the corporation’s to keep.
Nothing in that chain reaches into a human being. And nothing in that chain works in reverse. The corporation cannot turn human data into artificial data by collecting it. The collection changes who is holding the human data. It does not change what the data is, which is a record forever tethered to you and your life. The processing of human data may change its form. It does not change its substance. The human data the corporation has taken from you is what it was when you made it. The corporation has it. The corporation processes it. The corporation uses it. None of that makes it the corporation’s.
The law may say otherwise. The law has said otherwise about other things, and the saying did not make them so.
***
The corporation does something with the human data it has. It feeds the human data into its artificial machines, and the machines produce an output. AI outputs are not a record of a life. They are a probabilistic reconstruction of one. They are patterns the machine learned from millions of records of millions of lives, recombined into something the corporation can sell, at the cost of the land, labor, water, and human bodies on the other end of the extraction.
The artificial person built an artificial machine, fed it human data, ran the data through layers of computation, and produced artificial output. The output is an imitation of the human data. The imitation is then sold back to the people whose lives produced it, on terms set by the companies that took it.
***
As we have discussed, this has happened before, with cells. Henrietta Lacks[2] did not grow her cells to sell them. She grew them by being alive. The doctors who took them, the researchers who built an industry on them, and the law that decided her family had no claim to them all operated inside the same arrangement: a person’s body produces something of value, and the value belongs to whoever can extract it.
The arrangement was wrong then. It is wrong now.
Human data is taken the same way. It is taken without meaningful consent, processed, sold, and used to build a global industry. The institutions doing the taking have changed. The taking has not.
***
In the next edition, I want to make another argument from inside the AI industry’s own framework. It is an argument that follows from a way of thinking about the mind that many researchers and engineers in the AI industry already hold. That’s next.
***
[1]See Shoshana Zuboff, The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power (PublicAffairs 2019) (describing the translation of human experience into behavioral data in the context of advertising-driven platforms); B.S. Nayak & Nick Walton, Political Economy of Artificial Intelligence: Through the Prism of Liberal Capitalism (Palgrave Macmillan 2024) (analyzing AI’s applications in platform economies); Kate Crawford, Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence (Yale University Press 2021) (arguing that that AI is a material extractive industry built on mined minerals, exploited labor, and captured data, and that the costs are borne by the land and the people the industry treats as inputs).
