How easily can Russian propaganda fool AI models? A new benchmark finds out
New benchmark reveals how vulnerable AI models are to state-sponsored disinformation—and it's worse than you think.
Why it matters
As AI systems become critical infrastructure for information filtering, their susceptibility to coordinated propaganda campaigns is a governance and safety issue that boards and policy makers need to understand. This benchmark fills a gap in how we audit model robustness against adversarial information warfare.
The key facts
9 to knowInstitute of the Estonian Language released new benchmark
Measures AI model susceptibility to Russian propaganda
Published June 2026
Addresses AI safety/robustness in context of disinformation
Relevant to model evaluation and governance
Measures AI language model susceptibility to Russian propaganda
Published June 16, 2026
Addresses AI safety/security governance concern
No specific vulnerability percentages or model performance data provided in excerpt
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: The Institute of the Estonian Language has released a benchmark measuring how susceptible AI language models are to Russian propaganda.