The introduction of the superstore of vulnerabilities, as currently epitomized by Project Mythos, is shining a bright light on the potential vulnerabilities that organizations need to address in their digital infrastructure. Frontier AI models now have the context window and parameter depth to reason across the full complexity of a large network operating system codebase, surfacing interdependencies that experienced human researchers have historically missed. Currently, the volume of vulnerability disclosures has been accelerating for many years; leading-edge frontier models, however, can chain vulnerabilities together to create sophisticated exploits in hours, a task that once took years.