A recent cybersecurity scare involving Anthropic's advanced AI model, Mythos, has highlighted a burgeoning concern within the tech industry: the economics of AI scarcity. As Richard Waters points out in his commentary for the Financial Times, the increasing sophistication of frontier AI models is making access to these powerful tools a potentially critical factor in the global economy. The incident at Anthropic, though not fully detailed publicly, suggests that even leading AI developers are vulnerable to sophisticated cyber threats, raising questions about the security and control of these highly advanced systems.

The implication of such vulnerabilities is that the very availability and accessibility of cutting-edge AI could become a significant bottleneck. As these models become more capable and integrated into various sectors, from scientific research to financial markets, their scarcity could drive up costs and create a competitive disadvantage for those unable to secure access. This scarcity isn't just about computational power; it's about the proprietary knowledge, the specialized infrastructure, and the trained talent required to develop and deploy these models effectively.

Waters' commentary suggests a shift in the AI landscape, moving from an era of relative abundance in development to one where access to the most advanced capabilities might be tightly controlled and economically charged. This could lead to a concentration of power among a few key players who possess the resources to develop and maintain these frontier models. The cyber scare serves as a stark reminder that the infrastructure supporting AI is as crucial as the algorithms themselves, and any disruption can have far-reaching economic consequences. The future of AI development may well be shaped by who can afford and protect access to its most potent forms.