How to Detect and Measure the AI Dangers to Democracy

The paper presents a framework for assessing the risks AI poses to democratic processes, emphasizing that AI exacerbates existing democratic issues rather than introducing new ones. It leverages principal-agent theory to highlight accountability gaps in AI delegation and utilizes the NIST AI Risk Management Framework to establish measurable indicators for evaluating AI's impact on democracy. This approach aims to enhance institutional assessability and democratic control over AI, while acknowledging the limitations in current methodologies regarding the evaluation of harm and acceptable risk.