From Proposals to Enactment: The Procedural Bottleneck in AI Safety Regulation

Download PDF

Mansur Ali Khan, Mehmet Efe Akengin, Ahmad A Rushdi

Published: 23 Sept 2025, Last Modified: 23 Oct 2025RegML 2025 PosterEveryoneRevisionsBibTeXCC BY 4.0
Keywords: AI Governance, AI Safety, Regulatable AI, Legislation, AI Policy, Machine Learning, Policy, Legislation, Congress, AI Bills
TL;DR: Based on categorization of AI bill proposals it is found that US Congress members express interest in AI safety legislation but fail to pass effective laws, systemic barriers are identified, and actionable recommendations are offered.
Abstract: While large models advance rapidly, much AI safety legislation remains symbolic or stalled. While large models advance at near-exponential rates, AI safety legislation remains largely symbolic, stalled, or unrealized. Through a year-by-year analysis of AI breakthroughs, U.S. congressional policy proposals, and international legislative enactments, this study identifies a structural gap: the United States is not deficient in AI safety bill proposals but in legislative action, with only 4.23\% of U.S. AI bills reaching any terminal outcome. We analyze the trajectory of LLM releases versus global AI legislation, categorize U.S. Congressional AI bill sub-fields, "quantify enactment rates, identify specific procedural bottlenecks in the legislative process, and build a Logistic Regression model to identify that political and structural factors drive the low passage of AI safety legislation more than bill content. This study contributes four key advances: (1) a quantitative comparison of AI legislation versus LLM breakthroughs, (2) a comprehensive taxonomy of proposed and enacted policy sub-fields, (3) a dataset elucidating the causes of AI legislation failure, and (4) policy recommendations grounded in planned adaptation, preemptive enactment, and independent AI oversight. We demonstrate that without enactment, AI safety regulation remains inert, with a proportion of US AI bills that is less than expected. This paper underscores the urgent need for actionable AI safety policies in the United States.
Submission Number: 26