AI and Information Manipulation: Russia's Interference in the US Elections

Padalko's working paper is notable for its documentation of a phenomenon that most AI-disinformation analysis overlooks: Russian actors are not only using AI to generate and distribute disinformation content for human audiences, but are deliberately 'grooming' large language models with pro-Kremlin content — seeding training data and prompting environments to shape how AI systems respond to politically relevant queries. This two-track approach — targeting both human audiences and AI information systems — represents a significant doctrinal evolution.

The machine learning analysis methodology provides a more systematic basis for narrative tracking than most qualitative or semi-quantitative approaches. By identifying narrative clusters through ML rather than human coding, Padalko achieves a scale and consistency of analysis that is difficult to replicate manually. For practitioners building AI-assisted FIMI detection and monitoring systems, and for researchers assessing the state of the art in computational disinformation analysis, this working paper is a methodological contribution as well as a substantive intelligence document.