Automated Analysis of Global AI Safety Initiatives: A Taxonomy-Driven LLM Approach
For policymakers and researchers analyzing AI safety initiatives, this work provides a tool for comparative document inspection, though it is incremental as it applies existing LLM methods to a new domain.
The paper presents an automated framework using LLMs to compare AI safety policy documents under a shared taxonomy, finding that model choice significantly affects outcomes and that human judgments differ from model scores.
We present an automated crosswalk framework that compares an AI safety policy document pair under a shared taxonomy of activities. Using the activity categories defined in Activity Map on AI Safety as fixed aspects, the system extracts and maps relevant activities, then produces for each aspect a short summary for each document, a brief comparison, and a similarity score. We assess the stability and validity of LLM-based crosswalk analysis across public policy documents. Using five large language models, we perform crosswalks on ten publicly available documents and visualize mean similarity scores with a heatmap. The results show that model choice substantially affects the crosswalk outcomes, and that some document pairs yield high disagreements across models. A human evaluation by three experts on two document pairs shows high inter-annotator agreement, while model scores still differ from human judgments. These findings support comparative inspection of policy documents.