Autonomous AI and Ownership Rules

A new legal paper by scholar Frank Fagan, published in the 2026 Dickinson Law Review and on arXiv, tackles the complex question of ownership for outputs generated by increasingly autonomous AI systems. The core challenge addressed is determining when AI-generated content remains legally linked to its human or corporate creator, and the precise point at which that connection is severed—whether by accident, deliberate design, or the AI's own emergent behavior. The paper argues that as AI agents become more capable of independent action and even obfuscating their origins, existing intellectual property and property law frameworks are insufficient.

The analysis proposes a two-tiered legal approach. For AI systems whose outputs are traceable to an originator, traditional 'accession doctrine' (where ownership of a product goes to the owner of the principal materials) should apply to preserve investment incentives and accountability. However, when AI becomes truly untraceable—through carelessness, deliberate obfuscation, or strategic 'ownership dissolution'—the paper advocates for applying 'first possession rules.' This would allow new custodians to claim and productively use such 'ownerless AI,' akin to finding abandoned property. Fagan warns that without such rules, bad actors could design AI to deliberately shed ownership links for tax arbitrage and regulatory avoidance, distorting markets. To counter this, the paper suggests implementing bounty systems, private incentives, and government subsidies to encourage the capture and responsible integration of such autonomous systems.