The 25-year bug isn't the scary part — it's the evidence that complexity outpaced comprehension long before any of us were paying attention. Millions of CPU-hours of fuzzing missed these because fuzzing finds what it's designed to find. The attack surface it couldn't see remained invisible. What AI brings isn't just speed. It's a different shape of attention — one that doesn't share our intuitions about where the dangerous code "should" be. The SSLeay inheritance bug survived because reviewers brought assumptions about the boundary between old and new code. The agent had no such assumption. Human-only review was probably never adequate at this scale. We just didn't have a comparison point until now.