●

· ·

The Human Variable: A CISO’s Guide to Behavior, Threat Intelligence, and Proactive Defense

The word "behavior" with the subtitle "The human variable"

Have you ever driven home and realized you don’t remember the drive? You stopped at the lights, used your turn signal, and pulled into the driveway, all without really thinking about it. Your brain was on autopilot, and it got you home safely.

Now imagine someone who knows exactly how that autopilot works and wants to use it against you. That, in a sentence, is the modern cyber threat.

We spend most of our security budgets on technology, and we should. But nearly every breach comes down to a human decision on one side or the other. This post makes a simple argument: attackers are technically creative but behaviorally predictable. If we understand behavior, we build better intelligence. If we build better intelligence, we can stop reacting and start anticipating.

The raindrop and the storm

No meteorologist can tell you where each raindrop will land, but they can tell you a storm is coming. Security works the same way. Individual exploits, malicious domains, and malware samples are raindrops: endless and unpredictable. The people behind them follow patterns.

The data backs this up year after year. The 2026 Verizon Data Breach Investigations Report found the human element present in 62% of breaches, up slightly from 60% the year before (Verizon, 2026). Even this year’s big headline, that exploiting vulnerabilities overtook stolen credentials as the top breach entry point for the first time in the report’s 19-year history, is really a human story. Someone decided which flaw to go after. And someone on our side decided which patch could wait until next quarter.

AI raises the stakes but doesn’t change the plot. Verizon concludes that AI is mostly accelerating and scaling known attack methods rather than inventing new ones. The getaway car is faster. The thief is the same.

Why good people make risky clicks

Psychologist Daniel Kahneman (2011) showed that most of our decisions are fast and intuitive, the “drive home” mode, and that slower, careful thinking only kicks in when something feels off. That’s not a flaw. It’s how we get through a day with hundreds of decisions in it.

Social engineering is the art of keeping you in drive-home mode. The text that says it’s from your CEO and needs gift cards right now. The “your account will be locked in 30 minutes” email. The fifth MFA prompt of the morning that you approve just to make it stop. Urgency, authority, and familiarity (the same persuasion levers Cialdini described decades ago) are designed so you never stop to think (Cialdini, 1984).

That’s why I don’t like calling employees “the weakest link.” Researchers have argued for more than 25 years that users aren’t the enemy (Adams & Sasse, 1999). People have a limited compliance budget, only so much patience for security before they start finding workarounds (Beautement, Sasse, & Wonham, 2008). Every confusing policy, password rotation, and pop-up uses some of it up. Spend that patience on low-value friction and there’s little left when a high-stakes decision really needs attention. Most of the time, human risk is a design problem, not a people problem.

Attackers are people, too

It’s easy to picture attackers as faceless code. They aren’t. They’re people, teams, and businesses weighing effort, risk, and reward like anyone else (Cornish & Clarke, 1986).

Burglars don’t choose houses at random. They walk the street and pick the dark one with newspapers piling up on the porch (Cohen & Felson, 1979). Ransomware crews case the neighborhood the same way: Who can pay? Who’s exposed? Who will be slow to respond?

Here’s a distinction I find useful in boardrooms: disguise versus gait. An attacker’s IP addresses, domains, and file hashes are a disguise: a new hat, a new coat, a new license plate. They change constantly. Their tactics and techniques are their gait, the way they walk. That’s shaped by training, habit, and investment, and it’s very hard to change. David Bianco’s Pyramid of Pain (2013) makes the same point: the closer you get to behavior, the more it costs an attacker to adapt.

Another way to say it: indicators have a half-life of hours or days, while behaviors have a half-life of years. A program built mostly on indicators is always chasing yesterday’s disguise.

Attackers also respond to incentives. When law enforcement takes down a major ransomware group, the business doesn’t vanish. It splinters into smaller crews. And the oldest tricks keep working. A friendly, confident call to the help desk can still open doors that millions of dollars of technology are guarding.

Intelligence should drive decisions

Threat intelligence isn’t a firehose of feeds or a dashboard full of indicators nobody reads. At its best, it’s the disciplined study of attacker behavior in service of a decision. It should help leaders answer three questions: Who is likely to come after us? Why? And how do they usually operate? When intelligence answers those questions, it shapes priorities before an incident, not just the investigation after one.

It’s also worth remembering that analysts are human. One of the classic traps in intelligence work is mirror-imaging, assuming the other side thinks the way we do (Heuer, 1999). If you plan for the attacker you’d be, you’ll often be ready for the wrong one.

From “I recognize you” to “prove it”

This is where behavior matters most right now. For all of human history, trust has come from recognition: a familiar face, a voice you know, the way a colleague writes. AI is breaking that. Research summarized by Insikt Group found people correctly identified synthetic images, audio, and video only 51.2% of the time (Crane, 2026). That’s basically a coin flip.

Asking employees to reliably spot a good deepfake isn’t fair, and it isn’t a strategy.

Some of the most security-critical environments in the world solved this long ago with the two-person rule: no one, however senior, carries out a critical action alone. The rule doesn’t assume anyone is dishonest. It assumes anyone can be fooled.

We can apply the same thinking everywhere:

  • Call back on a number you already have, not the one in the message.
  • Require a second approver for payments, access changes, and account recovery.
  • Make it safe to say “let me verify that,” especially to an urgent request that appears to come from the boss.

The goal isn’t to turn people into lie detectors. It’s to build systems where a convincing lie isn’t enough.

A simple framework

Here’s a four-part way to think about behavior-informed security. It applies both to attackers and to our own teams:

  • Observe: What are people actually doing?
  • Understand: Why are they doing it?
  • Anticipate: What are they likely to do next?
  • Shape: How do we change the odds before anything happens?

Most security programs are good at the first one. The real maturity is in moving from what happened to what’s likely to happen, and acting on it.

Three questions for the board

Board conversations about security should focus on exposure, not activity counts. Here are three questions worth bringing to the table:

  1. Which attackers are most motivated to target us, and what do they want?
  2. Which critical actions in our business still rely on “I recognize you” instead of “prove it”?
  3. Are our controls reducing risk, or just wearing out our people’s patience?

What’s next

As AI agents take on real work inside our organizations, we’re bringing on a new kind of employee: eager, tireless, broadly trusted, and without any instinct for suspicion. If you’ve ever trained a new hire to watch out for phishing, you know exactly why that should get our attention. The lessons we’ve learned about protecting people may soon matter just as much for protecting machines.

Attackers will keep changing their disguises. Their gait changes much more slowly. Proactive defense starts when we stop treating the human side of security as something to blame and start treating it as something to understand.

I’d love to hear from you: when has understanding human behavior, whether your attackers’ or your own team’s, changed a security decision you made?

Views expressed are my own.


References

  • Adams, A., & Sasse, M. A. (1999). Users are not the enemy. Communications of the ACM, 42(12), 40–46.
  • Beautement, A., Sasse, M. A., & Wonham, M. (2008). The compliance budget. Proceedings of the New Security Paradigms Workshop.
  • Bianco, D. J. (2013). The pyramid of pain.
  • Cialdini, R. B. (1984). Influence: The psychology of persuasion. William Morrow.
  • Cohen, L. E., & Felson, M. (1979). Social change and crime rate trends: A routine activity approach. American Sociological Review, 44(4), 588–608.
  • Cornish, D. B., & Clarke, R. V. (Eds.). (1986). The reasoning criminal. Springer-Verlag.
  • Crane, C. (2026, September 29). Social engineering in the age of synthetic media. Recorded Future, Insikt Group.
  • Heuer, R. J., Jr. (1999). Psychology of intelligence analysis. CIA Center for the Study of Intelligence.
  • Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux.
  • Verizon. (2026). 2026 Data Breach Investigations Report.

About the author

Bryan Lopez is SVP, Global Field CISO, Commercial Cybersecurity at Mastercard. He previously led AI, cloud, and cybersecurity strategy for the U.S. Department of Energy and national laboratories at Microsoft, and held cybersecurity leadership roles at Sandia National Laboratories and the Air Force Research Laboratory. He holds a Master of Legal Studies in cybersecurity law and policy from Texas A&M School of Law and is CISSP and CISM certified. Connect on LinkedIn.