Junhai Liu (Renmin University of China – Law School) has posted The Legal Status of AI-Empowered One-Person Companies: Harnessing Benefits and Mitigating Risks—Focusing on China’s OPC Practice, Business-Entity Identification, and Digital Accountability on SSRN. Here is the abstract:
China’s national “AI+” initiative and local policies supporting one-person companies (OPCs) are accelerating the emergence of “one person + AI” and “lean team + AI” as new forms of entrepreneurship. Yet the term OPC is used inconsistently in Chinese policy practice: some documents employ it broadly to include companies, sole proprietorships, and individual industrial and commercial households, while others reserve it for a one-person limited liability company with a single natural-person shareholder. This Article argues that courts should distinguish the broad, policy-based concept of an AI+OPC from the narrow company-law concept. The latter should be confined to a duly registered company with a single shareholder that makes extensive use of artificial intelligence in its business operations. AI empowerment neither changes the company’s statutory organizational form nor turns AI that lacks legal personality into a shareholder, director, or independent bearer of liability. As a default rule, the legal consequences of AI-enabled business conduct should be attributed to the company that deploys, uses, and benefits from the system. The governance model through which AI participates in corporate decision-making and the type of business conduct performed through AI concern different questions—organizational authority and attribution of acts—and should not be conflated. The degree of AI autonomy should nevertheless be matched with corresponding duties of care, recordkeeping obligations, and accountability mechanisms. Judicial adjudication should also distinguish an entity-identification look-through inquiry from piercing the corporate veil. Courts should first identify the actual business entity by reference to registration and public disclosure, the outward appearance of contracting and performance, flows of funds, and algorithmic control; only then should they apply the liability regime attached by law to the identified entity, including independent corporate liability, the sole shareholder’s burden to prove separation of corporate and personal assets, the unlimited liability of a sole-proprietorship investor, or the liability of the operator of an individual industrial and commercial household. A sole proprietorship or individual industrial and commercial household may not obtain limited liability merely by using the label OPC or “one-person company”; conversely, a duly incorporated one-person company should not be recharacterized as either form merely because one natural person operates it with AI. Liability of foundation-model, interface, software, and other technology providers should turn on whether the relevant performance is a product, a service, or a hybrid arrangement and on the source of the defect, control, fault, and causation. Future judicial interpretations should adopt the principles of “priority to registered appearance, supplementation by actual business operations, statutory forms of liability, and cooperation in digital evidence,” supported by identity disclosure, an integrated “finance-algorithm-authority” evidentiary chain, human confirmation of material matters, algorithmic audits, and emergency takeover mechanisms. Whether highly autonomous AI should itself receive legal-person status remains a question for legislation, not present-day judicial interpretation.
Recommended!
To receive new posts from Legal Theory Blog by email, get a free subscription to Legal Theory Stack.
Lawrence Solum