# AI Models Risk Exposing Critical Vulnerabilities in Legal Systems Large language models could identify exploitable statutory gaps, potentially overwhelming current legislative frameworks. - **Source:** [www.theregister.com - Articles](https://brifing.si/source/2c34ea50-d341-4a10-925d-12c012a9ad10) - **Published:** 2026-11-02 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/a1a974c9-6794-498a-9418-487718d8d681 ## Legal Systems as Vulnerable Code Felix Ringe and Linus Folkerts argue that legal systems function similarly to computer code, containing inherent vulnerabilities. By treating written law as exploitable text, they suggest that large language models can identify inconsistencies at scale. This shift mirrors historical analogies from scholars like Lawrence Lessig, who previously explored the intersection of code and law. ## Automated Discovery of Statutory Gaps Ringe and Folkerts utilized large language models to uncover potential gaps within German law, specifically targeting the Occupational Diseases Regulation. They identified instances where statutory authority references were omitted, potentially invalidating specific provisions. These findings demonstrate how AI can systematically scan complex legislative frameworks to highlight technical errors that human reviewers might overlook. ## Risks to Legal System Stability The researchers warn that the proliferation of cheap legal reasoning via AI could jeopardize the functioning of modern legal systems. If adversarial actors use these tools to find exploits, the volume of challenges could overwhelm the judiciary. This creates a scenario where the speed of discovery outpaces the traditional statutory interpretation process used by courts. ## Limitations of the Code Analogy While the comparison between programming and law is useful, it remains imperfect. Courts do not function like deterministic runtime environments, as trial outcomes often involve non-linear human judgment. Despite these discontinuities, the core concern remains that AI-driven analysis could introduce the same chaos currently seen in the software security ecosystem into the legal world. ## Expanding AI Role in Courts Attorney Wendy Musell notes that AI adoption is already accelerating within the legal sector, with courts like the LA Superior Court testing AI as law clerks. These tools assist in researching and drafting opinions, forcing litigators to adapt. However, concerns persist regarding AI bias and the potential for inaccurate analysis of complex legal issues. ## Expert Perspectives on Methodology Professor Eric Goldman acknowledges the logic behind using AI to identify statutory gaps but emphasizes the necessity of human oversight. He notes that not all identified gaps carry legal significance or consequence. Legislatures often prioritize various political goals over technical perfection, meaning many discovered gaps may never be addressed by the governing bodies. ## Hardening the Legislative Process To mitigate these risks, Ringe and Folkerts suggest that lawmakers must harden the legislative process by leveraging advanced AI themselves. Goldman agrees that legislative counsel could benefit from AI tools during standard cleanup protocols. Integrating these systems could help identify drafting errors before laws are finalized, potentially reducing the surface area for future legal challenges. ## Future of Legal Infrastructure The intersection of AI and law represents a new frontier for systemic risk. As AI capabilities grow, the ability to weaponize statutory gaps will likely increase, necessitating a proactive approach to legislative drafting. Future stability will depend on whether institutions can effectively adopt these technologies to secure the foundations of social order against automated exploitation. ## Sources covering this story - https://www.theregister.com/ai-and-ml/2026/11/02/ai-could-break-legal-systems-by-finding-previously-hidden-loopholes/5307066 — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._