AI turns patches into working exploits in 30 minutes, and the 90-day disclosure window is the casualty

A security researcher warned that language models are speeding up vulnerability discovery and exploit creation, weakening the traditional 90-day disclosure model.

AI tools can help multiple researchers find the same bug faster and turn patches into working exploits more quickly, shrinking defenders’ response time.

Examples cited include flaws in an online store, React, and the Linux kernel, where patched code could be rapidly weaponized.

The concern is that disclosure embargoes and patch windows may no longer provide the safety buffer vendors assume, raising post-patch exploitation risk.