A Step Towards Modern Disinformation Detection: Novel Methods for Detecting LLM-Generated Text

New generative artificial intelligence (GenAI) technology could have devastating consequences on our democracy because it can be easily used to spread disinformation at scale while simultaneously personalizing propaganda to demographics or individuals. The threat is significant – Large-scale GenAI-based disinformation campaigns can sway public opinion, shape political events, or even compromise the integrity of elections. One method for defending against large-scale GenAI disinformation is to build tools for autonomously detecting AI-generated content. In this article, we evaluate state-of-the-art AI detection tools. Additionally, we propose a novel AI content detection method which demonstrates up to a 48% improvement in accuracy (over existing tools) for autonomously detecting AI-generated content.