AI Accountability and Liability

Artificial intelligence (AI) systems are increasingly influencing decision-making processes in critical domains, including healthcare, finance, and transportation. Determining who bears responsibility when AI systems cause harm or malfunction has therefore become a central challenge. This chapter addresses four core research objectives through a comprehensive review of existing literature on AI accountability and liability. The discussion begins with an examination of the doctrinal foundations that underpin current accountability frameworks. This addressed the first research objective. Particular attention is given to the European Union's AI Act and its accompanying directives, as well as comparative legal perspectives, including U.S. tort law and China’s regulatory approaches to autonomous technologies. The second component explores the economic dimensions of AI liability, focusing on compensation models designed to address the unique harms associated with AI‑enabled systems. Third, the chapter evaluates emerging legal and regulatory structures that govern AI liability, including ongoing debates around product liability. Several proposed models of responsibility are analysed, such as proportional and shared liability, risk‑based and sector‑specific approaches, and practical mechanisms such as contractual risk allocation and governance‑document frameworks. Existing accountability structures are critically assessed with respect to challenges posed by opaque “black-box” models, recognition of non‑physical harms, and the broader risk‑compensation dynamics within AI ecosystems. Fourth, concrete strategies and best practices were suggested to promote responsible AI development and deployment. This includes the establishment of global standards on AI responsibility and liability, domestication of AI laws to reflect the human rights and societal structures of countries, and the subjecting of all AI activities in all countries to risk assessments. Finally, the chapter advocates for a multi-stakeholder governance model that integrates legal oversight, industry self-regulation, and technological safeguards by providing a conceptual model reflecting the various theoretical perspectives to AI accountability and liability.