How Russian Disinformation Networks Are Poisoning AI Chatbots | SecureTalk with NewsGuard Experts

 In this eye-opening episode of Secure Talk, host Justin Beals investigates how foreign disinformation campaigns have evolved to target AI systems. NewsGuard researchers McKenzie Sadeghi and Isis Blachez reveal their groundbreaking investigation into the Moscow-based Pravda network that has successfully infiltrated Western AI chatbots, creating a new frontier in information warfare that bypasses human readers to directly corrupt the technology we rely on for information.

Key Insights:

• Leading AI chatbots repeated Russian disinformation 33% of the time when tested on key geopolitical topics
• The Pravda network has expanded from 50 to over 150 websites specifically designed to influence AI training data
• Modern disinformation operations now prioritize "LLM grooming" over targeting human readers
• Even when AI systems attempt to debunk false claims, they often cite unreliable sources, further legitimizing them
• Human oversight remains essential as these networks constantly evolve to evade automated detection

Notable Quotes: "Bad actors are targeting AI models to get their information to appear in those responses, but they're also weaponizing AI to produce disinformation at scale." - McKenzie Sadeghi

"What is changing is really the delivery of the disinformation... it's reaching a much larger audience and an audience that's not targeted as specifically because now practically anyone is using ChatGPT or other chatbots." - Isis Blachez

"This network does not invest any money or resources into spreading its content online... It's sort of serving as a centralized hub. And as a result, it appears very high not only in search results but also in chatbot responses." - McKenzie Sadeghi

Resources: 

• Sadeghi, M., & Blachez, I. (2025, March 6). A well-funded Moscow-based global disinformation network. NewsGuard Reality Check. https://www.newsguardrealitycheck.com/p/a-well-funded-moscow-based-global

#AISecurityThreats #Cybersecurity #Disinformation #InformationSecurity #TechSecurity #AIEthics #DataPoisoning #SecureTalk #InfoSec