tech
Europe's AI champion Mistral vulnerable to Russian disinformation, study finds
Open-source generative models are worse at removing false news than others, according to Estonian researchers
TL;DR
- Mistral AI, the EU's flagship AI initiative, is vulnerable to Russian disinformation campaigns, according to a new report.
- Researchers found Mistral's Mixtral 8x7B model can generate convincing fake news in multiple languages, including Russian.
- Concerns exist about the openness of Mistral's models and the lack of 'guardrails' to prevent misuse.
- Mistral AI argues that open access fosters innovation and allows for rapid identification and correction of potential risks.
- The company maintains its commitment to developing safe, responsible, and beneficial AI.