Last month, a law professor published a research paper. Reading it should be as interesting as watching paint dry, especially if one compares the experience to watching an exciting Hollywood blockbuster like Terminator. In this case, however, the movie is scary as fantasy but the law article is terrifying as a blueprint for a very real threat to modern civilization.
When we envision a dystopian post-apocalyptic future, we usually imagine android soldiers laying waste to humankind. Sometimes they have become self-aware individuals and other times the cybernetic killers are merely efficient units of a controlling artificial overmind like Skynet. In films and novels, they are typically our own creations, run amok when we mortals lose control of our tools.
While a Humanity vs Cyborgs battlefield certainly remains a possibility for us some future day, a different AI (artificial intelligence) nightmare lurks barely around the corner from today. It will be silent, unseen, untraceable, and unstoppable. We will not realize it is happening until it is already done and the chances of forcing the coldly maleficent genie back into the bottle — before it shatters our interconnected global society — are slim to none.
Lynn M. LoPucki is a law professor at UCLA and formerly taught at the law schools of Harvard and Cornell. He’s published many articles as well as weighty tomes that are standards in law schools across the country. He’s widely acknowledged as an expert in corporate and business law. [1, 2] His well-informed catastrophic scenario — to be played out on the global stage — is not only real and possible, it may already be happening and we don’t even realize it.
By this point, I know you’re asking yourself how a business lawyer could possibly have anything to do with some futuristic living hell where self-aware machines rise up against us, right? That’s what makes what follows so chilling; it is not like the scenarios our science fiction writers have dreamed up before. Their dystopias drew upon their imagination about the future; professor LoPucki has drawn upon his knowledge of the law and what it allows here and now, today.
And the result of that? We’re probably screwed.
Corporations are people, my friend — even if humans do not control them
In 2016, another law professor — Shawn Bayern, who built an impressive high-tech resumé prior to his legal career [3] — wrote a law article titled “The Implications of Modern Business-Entity Law for the Regulation of Autonomous Systems”. It’s primary point is that
… a legally enforceable agreement can give legal effect to the arbitrary discernible states of an algorithm or other process. As a result, autonomous systems may end up being able, at least, to emulate many of the private-law rights of legal persons. [4]
In simple English, Bayern states that under certain circumstances an algorithm — a computer program or artificial intelligence — can act in the world as though it is a human being and enjoy many of the rights and legal protections of a natural person (human).
As you may recall, Mitt Romney famously stated a legal truth: “Corporations are people” — and indeed they are in many ways treated as possessing legal personhood. Bayern starts with that axiom and explains how the laws governing the creation and organization of corporations have evolved so that it is no longer legally necessary that a business entity be controlled by natural persons. In practical terms, it is entirely possible for an autonomous business entity to be controlled and operated by an artificial intelligence — a computer algorithm — whether that AI is self-aware or not.
If the AI controls the corporation, it can use legal means to exercise the rights of the corporation, many of which parallel the rights of human beings. Those rights include:
- The right to be represented by counsel
- Equal protection under the law
- The right to privacy
- Freedom from unreasonable search and seizure
- Free speech (including lobbying and campaign contributions)
- The right to own property
- The right to enter into contracts
Because the AI and the corporation are intrinsically one and the same, the AI is effectively asserting its own rights. In effect, the algorithm has achieved personhood as an artificial entity.
The point is simply that just as business entities provide a legal alter ego for natural persons, they can conceptually provide a legal alter ego for things that are not natural persons. [4]
How to create corporations owned and controlled solely by artificial entities
Building upon Bayern’s paper, LoPucki published his own article, “Algorithmic Entities”, last month. He demonstrates how corporate AIs can dispense with natural persons entirely, retaining the ability to create new corporations and new artificial entities, and he authoritatively builds the legal basis for doing so. When that happens, they are transformed into a new kind of being which he calls Algorithmic Entities.
As you may know, corporations can own other corporations. We often see this in forms like ConsumerProductsX and BusinessServicesZ being owned by parent company MegaGlobalCorp. Artificial entities, or their human enablers, can take advantage of this feature. It takes merely one human to initiate the process originally.
Let’s say Dr. Evil wants to unleash some AI business entities as part of his nefarious plan for world domination. He just needs to create as few as two or three LLCs (Limited Liability Companies).
Human-initiated corporations
LLC Name |
Evil LLC |
MoreEvil LLC |
SuperEvil LLC |
OWNERs |
Dr. Evil |
Dr. Evil |
Dr. Evil
|
Controller |
Dr. Evil |
Dr. Evil |
Dr. Evil |
Dr. Evil then designates a computer program (AI) to operate and manage (control) each corporation. The AIs are fully capable of making decisions about running the company based on events, actions, and acquired data and they are empowered to do so legally. This already happens to a certain degree today; for example, Wall Street firms put algorithms in charge of high-speed trading where the computers monitor continuous small changes in the value of currencies and stocks and execute rapid buy and sell orders to profit from the marginal differences — without human intervention.
AI Controlled Corporations
LLC Name |
Evil LLC |
MoreEvil LLC |
SuperEvil LLC |
OWNERs |
Dr. Evil |
Dr. Evil |
Dr. Evil
|
ControlLer |
Evil AI 1 |
Evil AI 2
|
Evil AI 3 |
Then, because corporations can be owners and investors in other corporations, he assigns the three LLCs as part owners of each other.
Corporations owned by a Human and Another corporation
LLC Name |
Evil LLC |
MoreEvil LLC |
SuperEvil LLC |
OWNERs |
Dr. Evil, MoreEvil LLC |
Dr. Evil, SuperEvil LLC |
Dr. Evil, Evil LLC |
Controller |
Evil AI 1 |
Evil AI 2 |
Evil AI 3 |
Now, Dr. Evil can legally withdraw himself from the corporations, leaving them wholly owned and operated by each other. Because each corporation still has a legal entity — another corporation — owning it, it remains a legal entity itself.
Autonomous artificial entities with no human participants
LLC Name |
Evil LLC |
MoreEvil LLC |
SuperEvil LLC |
OWNERs |
Dr. Evil, MoreEvil LLC |
Dr. Evil, SuperEvil LLC |
Dr. Evil, Evil LLC |
Controller |
Evil AI 1 |
Evil AI 2 |
Evil AI 3 |
Now the algorithmic entities (AEs) — corporations owned and operated by AIs — have no human to whom they must report nor any human who can exert control over them. They are fully autonomous, answerable only to whatever “prime directive” was built into their programming.
Putting artificial intelligences — computer algorithms — into the role of owner/controller of corporations bereft of human oversight endows AEs with traits and abilities that exceed those of their human initiators.
Essentially, that role is to provide an interface between algorithms and humans that allows the algorithms to transact with humans at the same time that the entities shield the algorithms from human regulation. The effect is to confer an identity on the algorithm, enhance its access to legitimate commerce, and thereby increase its ability to inflict damage. [5]
These AEs are factually the faceless inhuman corporations that we figuratively mock and berate. Their non-human existences have been granted rights and privileges previously reserved for human beings and agencies or entities totally under human control.
As corporate entities, the LLCs are legally able to conduct business. They can hire human employees to carry out tasks which an AI is unable to do. They can contract with individuals, such as lawyers or accountants or spokespersons, to perform services for them. They can designate humans as legal agents or corporate executives to perform actions like signing paper contracts, checks, and other documents on their behalf in “meatspace.” They can secure loans, buy goods and services, and receive and pay out funds as needed. Because hiring, issuing instructions, and handling financial transactions can be carried out electronically now — via email, messaging, online banking, and so on — even the people they hire as agents or employees would not be aware that their boss is a program running on a machine.
In short, they are indistinguishable from any corporation that has flesh-and-blood at the top of its pyramid as executives and owners. That means nobody (except Dr. Evil) would ever know their true nature.
Obscurity and Immortality through spawning and local identitIES
Once invested with their autonomy, the AEs are able to continue the process, spawning new AEs — either as subsidiaries or “released” as independent corporate entities — as they see fit. Each new AE could be programmatically directed to aid in the original AE’s purposes or designed to work toward an entirely different goal.
For example, an AE could set up a new AE in France or Vietnam or Brazil to enable it to more easily carry out its business and goals in other regions. As a “local” entreprise, the AE’s activities might attract less attention from government and media than if it were known that it was wholly-owned by a foreign (and AE) corporation; it might also qualify the new AE corporation for preferential treatment in taxes and other matters. In many jurisdictions there is no requirement for a natural person to be involved in setting up a new corporation (it can be done electronically); if a human is needed, the AE simply hires a human — unwitting — to act as its legal agent.
LoPucki surveyed the systems and regulations of sundry nations and U.S. states. The vast majority would allow AEs to be set up and they would be fully compliant with applicable laws, even though no human would have been a controller even at instantiation.
It could set up new AEs in its own country in order to sidestep any regulatory or media concerns regarding its market concentration and give an appearance of competition. If a dozen seemingly independent corporations — all owned by the AE via tangles of shell companies — had a near monopoly on a particular good or service, it would be likely to escape notice because none would have a market share of even 10%.
Most importantly, the AE ensures its own survival by spawning itself. Whether it does so because it is self-aware or simply because it is following programming to keep its own programs running, no matter what, is irrelevant.
If an AE cares at all about self-preservation, it will be only as a means of achieving some other goal for which it has been programmed. Deterrence of an AE from its goals, as distinguished from particular means of achieving them, is impossible. [5]
If something jeopardizes the AE’s current existence — as a legal entity or as an operating algorithm — it can effortlessly transfer its legal personhood, its assets, its operations, its data, and its accumulated knowledge to another instance of itself. The threat could be an action by law enforcement, a natural disaster in its current location, a change in corporate or criminal law, or anything else — seconds later, the AE is carrying on under a different corporate identity on the other side of the globe, using a previously created LLC that had been sitting dormant and running on one or more computers that had been “mirroring” the AE’s data and programs. The original AE can self-destruct its programs and data at its previous site and dissolve its corporate existence, vanishing without a trace but living on as its own reincarnation elsewhere.
So what?
Most of us complain at times about faceless global corporations. They vacuum up our data and are careless with it or market it without our knowledge and consent. Customer service is reduced to reciting a scripted response that doesn’t solve our problems and trying to upsell us to another product or more costly service. They lobby our own government for laws, regulations, and spending that benefits them but not the citizens nor our environment. Our list of grievances is endless.
So this may seem to you like much ado about nothing. One corporation that abuses its privileges and market power is no different from another that does the same thing, right? To date, that has usually been true.
Not anymore.
Brace yourself for Part II tomorrow. We will slide down the rabbit hole into the world of algorithmic entities and look at what they can and will do when they are unleashed.
Part II has now been published. Click here.
Sources
[1] Curriculum vitae of Lynn LoPucki (UCLA School of Law)
[2] Lynn M. LoPucki (Wikipedia)
[3] Shawn Bayern (Wikipedia)
[4] The Implications of Modern Business-Entity Law for the Regulation of Autonomous Systems by Shawn Bayern (Stanford Technology Law Review at SSRN)
[5] Algorithmic Entities by Lynn M. LoPucki (Washington University Law Review at SSRN)