Mistral CEO 警告法国不要让 Anthropic 的 Mythos 扫描军事代码
mistral 首席执行官 arthur mensch 向一个法国委员会发出警告，不要让 anthropic 的 mythos ai 扫描军事代码库，引用了不可逆转的网络安全依赖风险。
mistral ceo arthur mensch warned the french commission of inquiry that letting anthropic's mythos scan military codebases creates an irreversible dependency, as modern ai models can orchestrate attacks and find exploits. he emphasized this risk extends beyond us systems to models built by mistral or chinese developers. despite this, the eu continues negotiating with openai and anthropic for early access to advanced cybersecurity models. mensch affirmed mistral's independence, noting us investors hold under 30% and an ipo is planned.