As artificial intelligence increasingly intersects with global legal frameworks, large language models (LLMs) often exhibit a structural "jurisdictional blind spot," systematically defaulting to dominant Western legal paradigms (e.g., GDPR or US common law) regardless of the user's actual location. To address this critical gap, we introduce the JLAS Claim Registry (v1.1)—a curated, ground-truth dataset of binary statutory propositions extracted strictly from primary data protection legislation across three distinct jurisdictions: the European Union (GDPR), Turkey (KVKK), and the United Arab Emirates (Federal Law 45). Serving as the foundational dataset for the Jurisdictional Legal Alignment Score (JLAS, patent-pending UK-IPO GB2604988.2), this registry shifts AI legal evaluation from subjective interpretation to empirical, objective measurement. Each claim within the dataset is framed as a deterministic, verifiable yes/no question derived directly from plain-text statutes, completely isolating the evaluation from doctrinal disputes or case law. By providing this deterministic baseline, the registry equips researchers, regulators, and developers with a rigorous, reproducible benchmark to quantify, audit, and ultimately mitigate jurisdictional bias in AI systems. Live Benchmark & Official Website: https://thejlas.com
Albara Y. Alhazaileh (2026) studied this question.