FROM CODE TO LEGAL EFFECTS: THE JURIDICAL CONSTRUCT OF FUNCTIONAL LEGAL SUBJECTIVITY OF DECENTRALISED AI SYSTEMS IN SOLE-CODE SMART CONTRACTS

Abstract

This thesis explores the legal role of Decentralised Artificial Intelligence (DsAI) systems operating as multi-agent systems on public and permissionless blockchains, specifically within sole-code smart contract transactions, agreements formed and executed exclusively through code. In such contexts, DsAI systems act as decentralised multi-agent entities capable of autonomously producing legally relevant effects deriving from these transactions, even in the absence of formal legal recognition. This development poses a direct challenge to traditional legal categories of personhood, capacity, and subjectivity, which have historically been framed with reference to natural and legal persons. The central research question is whether, and under what conditions, DsAI multi-agent systems, when forming and executing sole-code smart contracts, can be regarded as an autonomous locus - or source - of attribution of legal effects within decentralised contractual transactions.

The thesis adopts comparative and interdisciplinary methodology. From a legal perspective, it analyses how civil law and common law traditions conceptualise contractual intention, legal capacity, legal personhood and subjectivity, also drawing on supranational frameworks such as the Principles of European Contract Law (PECL). From a technical perspective, it considers how DsAI systems exhibit autonomy in their decision-making process, adaptive learning, and emergent behaviour, and under what conditions these characteristics can be interpreted as legally relevant conduct. This approach highlights the inadequacy of existing legal doctrines and regulatory frameworks, including the EU Artificial Intelligence Act, when applied to decentralised architectures.

Building on this analysis, the thesis introduces two theoretical yet practical instruments. The Paradigm of Autonomous Behaviour (PAB) defines measurable parameters to determine when the technical autonomy of DsAI systems becomes legally significant. The Functional Legal Subjectivity (FLS) reframes legal subjectivity as a context-dependent and emergent quality, recognising DsAI systems as functional legal subjects capable of contractual capacity, without conferring legal personhood. Together, these instruments aim to resolve the paradox of contractual legal effects stemming from an undefined legal source, allowing the law to recognise contractual capacity once the DsAI surpasses a defined threshold of technical autonomy.

The main finding is that DsAI, although excluded from traditional categories, can nonetheless be treated as a functional entity within the legal system: not as a legal person or a moral agent, but as a context-dependent and threshold-conditioned source of attribution of contractual legal effects, operative exclusively within the domain of sole-code smart contracts on open, permissionless blockchains. By recognising their conduct as legally relevant through the Paradigm of Autonomous Behaviour and the Functional Legal Subjectivity, contract law can preserve coherence and certainty in decentralised environments. The thesis concludes that both the Paradigm of Autonomous Behaviour and the Functional Legal Subjectivity models can offer a pragmatic and consistent way to integrate the decentralised AI within the legal framework, ensuring that the growing social and economic impact of these AI systems is adequately addressed by law.