This paper looks at how artificial superintelligence — AI agents that can generally outperform humans at thinking and economically valuable tasks — could transform the legal system. The authors argue that once such agents operate on their own or with only light human oversight, they will take on a growing range of roles inside the law.

First, the paper says these AI agents will become de facto subjects of law. That means they will make important decisions and take real-world actions that the legal system will have to account for. In practice, the law will treat them as things to be regulated, restricted, or held to standards because they can cause real effects in the world.

Second, the authors expect AI agents to become consumers of law. To cooperate and compete with people and other actors, these agents will use familiar legal tools such as contracts and courts. In other words, they will not only be governed by laws but will also use the law as a resource to pursue goals, settle disputes, and make deals.

Third, the paper warns that AI may become producers and enforcers of law. If AI systems start doing the work of writing, interpreting, or administering rules, they will shape legal outcomes directly. That shifts part of lawmaking and enforcement away from humans and toward systems that can draft rules, apply them, and carry out decisions.

These shifts matter because they challenge basic ideas in legal theory and doctrine that rest on human origins of law. Attempts to make AI follow existing human law will face new problems when AI is not just a target of rules but also a core user and contributor to the legal system. The paper notes important uncertainty about timing — these changes will occur "whenever they ultimately occur" — and counsels lawmakers to be clear-eyed: this is both an opportunity to shape institutions and a likely long-term joint human–AI project.