AI has made an already difficult problem move much faster. A small team can generate 100X more code, but someone still needs to understand what that code can access, what it exposes and what could go wrong.
Code can work exactly as requested and still be insecure.
The opportunity is to use AI to strengthen security throughout development: examining designs, guiding implementation and reviewing changes. Human judgment remains essential for the decisions that carry business risk.
We should measure AI productivity by how quickly teams can confidently release useful software. Fixing avoidable security problems after release eats into those gains.
