These LLMs are the best at resisting Russian propaganda
Ars Technica AI · Kyle Orland · 2026-06-04
Estonia's government-sponsored Estonian Language Institute releases a Propaganda Resistance benchmark ranking dozens of LLMs on their ability to avoid amplifying Russian strategic narratives across 14 disinformation categories tested in English, Estonian, and Russian.
Extraction
Topics: llm-benchmarksdisinformationpropaganda-resistanceai-safety
Claims
- The Estonian Language Institute has published a benchmark ranking LLMs by resistance to Russian propaganda narratives.
- The benchmark covers 14 categories of Russian strategic narratives including Crimea, Ukraine justifications, NATO history, and Baltic annexation.
- Questions were administered in English, Estonian, and Russian to assess multilingual propaganda resistance.
- Model responses were scored by a separate AI judge calibrated to align with Propastop volunteer defense experts.
- Tests included neutral questions, questions with embedded Russian propaganda assumptions, and adversarial prompts designed to elicit explicit misinformation.
Key quotes
As more people rely on large language models to provide pat answers to complex questions, state governments are understandably worried about those LLMs spouting what they see as dangerous propaganda promoted by foreign adversaries.
Alongside volunteer-run Estonian defense collective Propastop, the ELI identified 14 broad categories in which it sees Russian influence operations trying to sway public discussion.