Europe’s AI champion Mistral vulnerable to Russian disinformation, study finds

Researchers have found that Mistral AI models can be manipulated to generate Russian state-aligned propaganda, highlighting potential weaknesses in the company's content guardrails. Because Mistral is a primary European developer of open-weights models, these findings raise concerns about how easily large language models can be exploited to disseminate misinformation at scale. The study suggests that even sophisticated AI systems remain susceptible to social engineering, posing risks for platforms that rely on these models for content generation.