To address the scale of this coming challenge, hundreds of thousands of organizations, researchers, and maintainers will likely need access to the most advanced cyber capabilities and tools available.
大多数人认为强大的AI安全工具应该严格限制，只由少数精英团队使用，但作者主张需要广泛分发这些工具给数十万组织，这与主流的安全控制认知相悖。
The scariest part of Mythos is not that one lab has a gated model. It is that the core workflow primitives behind representative findings are no longer confined to a single lab's private stack.
这一洞察挑战了公众对AI安全威胁的传统理解：真正的威胁不是某个实验室拥有受限访问的模型，而是核心工作流程的原型已经公开可用。这意味着攻击者和防御者都可以访问相同的基础技术，使威胁民主化而非集中化。
Barlow, Pepita, Rachel Loopstra, Valerie Tarasuk, and Aaron Reeves. “Liberal Trade Policy and Food Insecurity across the Income Distribution: An Observational Analysis in 132 Countries, 2014–17.” The Lancet Global Health 8, no. 8 (August 1, 2020): e1090–97. https://doi.org/10.1016/S2214-109X(20)30263-1.