“The architect of the invasion” vs. “a symbol of resilience”: How AI chatbots frame Zelenskyy, Putin and the Russia–Ukraine war

Abstract

This qualitative content analysis conducted in November 2024 examines how four AI chatbots – OpenAI’s ChatGPT, Microsoft’s Copilot, Google’s Gemini, and Anthropic’s Claude – frame the Russia–Ukraine war. The chatbots generally utilized similar emphasis, tonality, and moral judgements as most US and European news organizations, with Russia framed as an “aggressor” and Ukraine as a victim of the aggression. These framing similarities were even more pronounced in outputs for Russian and Ukrainian identity-specific prompts. However, AI chatbots had a hard time contextualizing Russian government propaganda. Thus, AI chatbots are unreliable narrators, prone to bias and a superficial understanding of complex geopolitical realities. Until these fundamental issues of reliability and bias are resolved, AI chatbots as news sources should be treated with significant skepticism.