Models Shrink Zero‑Day Value

Powerful generative models that automate vulnerability discovery and rapid patch suggestion may drastically shorten the exploitable lifetime of a zero‑day, reducing its expected payoff for attackers. If true, defenders who deploy model‑driven scanning and remediation could make most offensive research unprofitable, creating a new cyber equilibrium where mass investment in hacking no longer pays. — This reframes cybersecurity policy and military planning: AI could shift the offense–defense balance toward defense, altering deterrence calculations, procurement priorities, and norms around model access and trust.