AI Could Soon Lead Businesses – Ensuring Compliance with Legal Regulations

Daniel Gervais from Vanderbilt University and John Nay from Stanford University collaborated on this news article.

Tennessee, October 29 (Discussion) Only “individuals” can interact with the legal system – for example, by signing agreements not to file lawsuits.

There are two main classes of persons: human beings called “natural persons” and creatures of the law called “artificial persons.” These include corporations, non-profit organizations, and limited liability companies (LLCs).

Until now, the purpose of artificial persons has been to help people achieve certain goals.

For example, people can pool their assets in a company and limit their liability towards customers or other people dealing with the company. But a new kind of artificial human is poised to enter the scene – artificial intelligence systems, and they may not serve human interests.

As researchers studying AI and the law, we believe that this moment presents a significant challenge to the legal system: how to regulate AI within existing legal frameworks to reduce unwanted behavior and how to provide legal accountability for the autonomous actions of AIs.

One solution is to teach AIs to be law-abiding entities.

This is far from a philosophical question. LLC laws in several US states do not require people to supervise the LLC. In fact, in some states it is possible to form an LLC without a human owner or “member” – for example, in cases where all parties are deceased.

While lawmakers probably didn’t have AI in mind when they drafted the LLC laws, the possibility of zero-member LLCs opens the door to AI-run LLCs.

Many small and large business functions have already been partially delegated to AI, including financial functions, human resources, and network management, to name just three.

AIs can now perform many tasks as well as humans. For example, AIs can read medical X-rays and perform other medical tasks, as well as perform tasks that require legal justification. This process is likely to accelerate due to innovation and economic benefits.

A different person

Humans have sometimes included non-human entities such as animals, lakes and rivers, and corporations as legal subjects. Although in some cases these communities can be held accountable for their actions, the law only allows people to participate fully in the justice system.

One major obstacle to the full access of non-human entities to the legal system has been the role of language as a uniquely human invention and an important part of the legal system. Language enables people to understand the norms and institutions that make up the legal framework. But humans are no longer the only ones who use human language.

Recent advances in AI’s ability to understand human language open up its potential to interact with the legal system. Artificial intelligence has demonstrated competence in a variety of legal tasks, such as tax law, lobbying, contract drafting, and legal reasoning.

An LLC formed in a jurisdiction that allows it to operate without human members could trade in digital currencies calculated on blockchains, allowing the AI running the LLC to operate autonomously and in a decentralized manner, making it challenging to regulate.

Under a legal principle known as the doctrine of domestic affairs, even if only one U.S. state allowed AI-powered LLCs, that entity could operate nationwide — and potentially globally. This is because the courts look to the rules for the internal affairs of the general meeting based on the law of the state of incorporation.

Therefore, we believe that the best way forward is to align AI with existing laws, rather than creating separate rules for AI. Additional laws can be imposed on man-made substances, but artificial intelligence should at least be subject to all laws subject to humans.

Building law into artificial intelligence

We recommend a research direction to integrate the law into AI agents to ensure compliance with legal standards. Researchers could train artificial intelligence systems to learn methods to internalize the spirit of the law.

The training would use knowledge generated by legal processes and tools of the law, including legislative methods, statutory interpretation, contract drafting, application of legal standards and legal reasoning.

In addition to incorporating the law into AI agents, researchers can develop AI compliance agents—AIs designed to help an organization automatically comply with the law. These specialized AI systems would provide third-party legal safeguards.

Researchers can develop AI legality by fine-tuning large language models of how to perform legal tasks labeled with supervised learning. Another approach is reinforcement learning, which uses feedback to tell the AI whether it’s doing a good or bad job—in this case, lawyers interact with language models. Legal experts could design prompting systems—ways to interact with the language model—to elicit better responses from language models that are more consistent with legal standards.

Law-abiding (artificial) entrepreneurs

If the LLC was run by an AI, it would have to follow the same laws as any other LLC, and the courts could order it to pay damages or stop doing something by issuing an injunction. An AI tasked with running an LLC and, among other things, maintaining proper business insurance would have an incentive to understand and comply with applicable laws. The minimum requirement for company liability insurance is a standard requirement that most companies impose on each other in order to enter into a business relationship.

The incentives to form AI-powered LLCs are there. Fortunately, we believe that it is possible and desirable to work on embedding the law – which until now has been human law – into AI and AI-powered automated compliance guardrails. (Talk) GRS GRS