AI & Society
Cybersecurity Is a Human Problem First
Most breaches are not a failure of technology. They are a decision somebody made, or did not make, months earlier.
When people picture a security failure, they picture a brilliant attacker and a broken system. Somebody in a dark room finding a hole nobody could have seen.
That happens. It is not the usual story.
The usual story is quieter. A password that got reused. An access permission nobody removed when someone changed roles. A warning that fired so often everybody learned to ignore it. A deadline that made "we will fix the security part later" sound reasonable.
Those are human decisions. The technology just carried them out.
Incentives decide more than firewalls
If you want to know how secure an organization is, do not start with the tools. Start with what people get rewarded for.
If engineers are rewarded for shipping fast and never for shipping safely, they will ship fast. If the person who raises a concern gets treated as the person slowing everybody down, people will stop raising concerns. If reporting a mistake gets you blamed, mistakes will go unreported until they are too big to hide.
None of that shows up in a security audit. All of it shows up in a breach.
Some of my earliest work in this area was at the Federal Reserve Bank of New York, from 2006 to 2008. I helped design products for cybersecurity and incident response, including work supporting the National Incident Response Team. The job was taking complex security and operational data and turning it into something people could act on quickly, under pressure, in a highly regulated environment. What stayed with me is that the technology was only half of it. The other half was whether the people in the room could see clearly and decide fast.
People are not the weakest link
You will hear security people say that humans are the weakest link. I understand why they say it. I also think it gets the relationship backward.
If a system can be brought down by one tired person clicking one convincing email, the problem is not the tired person. The problem is a design that put that much weight on one moment of attention.
Good security assumes people will be tired, distracted, rushed, and kind. It is built so that a single human mistake is survivable. That is a design choice, and it is made long before anyone clicks anything.
AI raises the stakes on all of this
AI has made it much cheaper to write a convincing message, to imitate a voice, and to try thousands of variations of an attack. The emails that used to give themselves away with bad grammar do not anymore.
That does not change the core of the argument. It sharpens it. When deception gets cheap, the protection has to come from habits and systems that do not depend on spotting the deception. Verification steps for money and access. Clear ways to say "I am not sure this is real" without feeling foolish. Leaders who model that themselves.
What leaders can actually do
Make it safe to report. The fastest way to find problems is to stop punishing the people who find them.
Make security part of the work, not a gate at the end. If it is a final checkpoint, it will always feel like an obstacle.
Ask the uncomfortable question in planning. What is the worst thing someone could do with this feature? Say it out loud before launch.
Own the tradeoffs. Sometimes you will choose speed over safety. That can be a legitimate business decision. It should be a decision someone made on purpose, not an accident nobody noticed.
It starts with how we treat each other
I have spent a lot of my life paying close attention in rooms, noticing who is about to be the problem before anybody says anything. In security, that instinct matters. But the more useful skill is building a room where people tell you the problem themselves, early, without fear.
That is a culture question. It always was.
If your leadership team wants to talk about cyber risk in plain language, I speak and advise on exactly this.
