Legal Challenges Related to Contractual Negotiations via AI Technologies: Comparative Analytical Study

Background: Contractual negotiations conducted by artificial intelligence (AI) systems raise profound legal challenges, most notably the question of allocating civil liability for damages caused by their errors. This study, employing a comparative analytical methodology, reveals a significant regulatory gap in Arab jurisdictions—particularly Egypt and the United Arab Emirates—where legislation lacks explicit provisions governing such liability. In contrast, recent European Union initiatives, including the risk-based approach of the AI Act and the emerging framework of the AI Liability Directive, place primary emphasis on the accountability of developers and operators. Against this backdrop, the paper advocates for the development of a specialised Arab legal framework that draws inspiration from comparative models while preserving local specificities. Such a framework should include: a precise legal definition of intelligent systems, concrete evidentiary mechanisms for fault attribution and liability distribution, the establishment of a dedicated supervisory authority, and the strengthening of insurance mechanisms as complementary safeguards. Methods: This study employs a comparative analytical method to examine civil liability for AI errors in contractual negotiations, focusing on tort and contractual theories under Egyptian and Emirati law, and contrasting them with recent EU developments—particularly the AI Liability Directive and the Data Act, which provide clearer guidance than the AI Act. Results and conclusions: The comparative analysis yields three main results. First, there is a clear regulatory gap in Arab jurisdictions, which continue to rely on general civil law provisions without specialised rules for AI. Second, doctrinal differences between strict liability in the EU and the broader remedial approach in Arab systems complicate any direct transposition of European models. Third, evidentiary challenges remain central in both systems, as establishing fault and causation in AI-related harm is inherently complex.