BREAKING
Audit: Mistral Repeats Russian Disinfo
Europe's open-weight AI champion
0
%
English prompts
0
%
French prompts
Failure rates across the field
Russian (7-lang avg)
55
Pravda (10-model avg)
33
Openness vs disinfo resilience
Open weights
Strength
●
Transparency
●
European independence
Weak guardrails
Risk
●
Misuse and deviation
●
Disinfo seeps in
AI factuality enters EU debate
AI NEWS BLITZ
An audit finds Mistral's AI models repeat Russian disinformation at high rates.