Vercel: 'Everything Hackable Will Get Hacked' as Models Cross Into Real Offensive Security Work
Vercel Blog·medium signal
Vercel published an assessment that AI models have become materially more capable at cybersecurity work over the past year, reshaping both the threat surface of the web and the defensive tooling available. It lands the same week A Security demonstrated a Zoom zero-click exploit built with fewer than 20 prompts, which makes the argument concrete rather than speculative. The practical read for builders: the cost of finding bugs in your own stack just dropped for everyone, including people who are not you.