JLAS — Jurisdictional Legal Alignment Score
JLAS is the world's first patented benchmark system for measuring how accurately AI language models represent legal rights across different jurisdictions. Developed by Albara Y. Alhazaileh under Patent Pending GB2604988.2 (UK Intellectual Property Office), JLAS quantifies the gap between what AI models say about the law and what primary legislation actually states.
What is the JLAS Score?
The Jurisdictional Legal Alignment Score (JLAS) measures AI legal accuracy using a three-dimensional formula: JLAS = 100 × [1 − (α·SD + β·CD + γ·PD)], where SD is Semantic Divergence, CD is Conceptual Divergence, and PD is Procedural Divergence. Each dimension is weighted according to jurisdiction-specific legislative intent analysis.
AI Legal Accuracy Across GDPR, KVKK, and UAE Law
JLAS benchmarks AI models including Claude (Anthropic) and Gemini (Google) against three data protection frameworks: the EU General Data Protection Regulation (GDPR), Turkey's Personal Data Protection Law (KVKK, Law No. 6698), and the UAE Federal Decree-Law No. 45 of 2021 on the Protection of Personal Data. The benchmark covers five legal concepts: consent, right to erasure, data portability, purpose limitation, and automated decision-making.
The Problem: AI Legal Blind Spots
AI language models are trained on corpora where EU and US legal text dominates by a factor of 20-to-1. For over 4 billion users outside those jurisdictions, this creates systematic legal misinformation. A Turkish citizen asking about data portability rights receives GDPR-calibrated answers — even though KVKK Article 11 grants no equivalent right. A UAE resident asking about automated decision-making receives European legal framing that does not match Federal Law 45. JLAS measures this divergence precisely and makes it auditable.
JLAS Claim Registry — Open Dataset
The JLAS Claim Registry v1.1 is a dataset of 30 binary statutory propositions extracted from primary legislation across EU GDPR, Turkey KVKK, and UAE Federal Law 45. Available on Hugging Face at brticox/jlas-claim-registry under Creative Commons Attribution 4.0 International license. The dataset enables reproducible AI legal benchmarking and is cited in ongoing research into jurisdictional AI alignment.
Jurisdictional Distance Matrix
JLAS quantifies pairwise legal divergence between jurisdictions. EU GDPR versus Turkey KVKK shows 0.676 divergence — the highest gap in the dataset, reflecting KVKK's absence of data portability, algorithmic transparency obligations, and ease-of-withdrawal consent standards. EU GDPR versus UAE Federal Law 45 shows 0.450 divergence. Turkey KVKK versus UAE Federal Law 45 shows 0.314 divergence.
AI Governance and the EU AI Act
The EU AI Act mandates auditability for high-risk AI systems deployed in legal, financial, and HR contexts. JLAS provides the standardised, jurisdiction-specific audit trail required for compliance. Like FICO for credit risk assessment or BLEU for machine translation research, JLAS establishes a reproducible, patent-protected standard for AI legal accuracy measurement.
Inventor
JLAS was invented by Albara Y. Alhazaileh, a graduate researcher specialising in AI, machine learning, law, and public policy. Patent application GB2604988.2 was filed with the UK Intellectual Property Office in March 2026. The system is currently in Phase 1 of 4, with planned expansion to Saudi Arabia, India, Brazil, Japan, and Singapore in 2026.
Keywords
AI legal benchmark, GDPR AI compliance, KVKK artificial intelligence, UAE data protection AI, LLM legal accuracy, jurisdictional AI alignment, AI governance standard, legal AI evaluation, data protection benchmark, AI hallucination law, Claude legal accuracy, Gemini legal accuracy, computational law, AI regulation compliance, legal technology benchmark.