Microsoft's monthly update, known as Patch Tuesday, is drawing attention. This time the company fixed a record 570 security flaws across its products. What stands out is that Microsoft says AI helped find many of them. AI can now scan for weak spots that humans often miss. Behind each monthly Windows update, the work of defense itself is starting to change.

Why now? Finding flaws means reading huge amounts of code and spotting risky patterns. This is exactly what modern large language models are good at. Their ability to guess what code means and flag dangerous lines has become practical. But attackers can use the same tools. If defenders get faster, so do attackers. That is the heart of this story.

OpenAI has recreated this tug-of-war between attack and defense in-house. It built a super-hacker AI called GPT-Red. The goal is to let it attack the company's main models and toughen their defenses. By making AIs fight each other, a method called self-play, they can block attack ideas humans might never imagine. The stronger the sparring partner, the stronger the real model becomes.

The defense story also connects to jobs. Anthropic and others launched a venture called Ode, betting that a few engineers can do the work of an army of consultants. It hints that professional service work may shrink through AI. From a long-term view, the key question is where the line falls between tasks AI can replace and judgment only humans can make. It may be worth listing which parts of your own work AI could handle.

There are also signs of what comes next. Vint Cerf, a father of TCP/IP (the basic rules of the internet), is working on a standard to identify AI agents. In an age where AI moves around the net on its own, we need a way to prove who an agent is and who it works for. Just as humans have driver's licenses, agents may need IDs too. The stronger this foundation, the more safely we can hand work to AI. It is a field worth watching.