I can remember this event vividly. It was genuinely frightening at the time. Major parts of the Internet kept going dark, and no one seemed to know why. So while my first instinct is to dismiss today’s doom-and-gloom warnings about AI destroying the world as science fiction, it came back to me that a milder version of exactly this kind of scenario actually happened—in 1988, with no AI involved at all.

Remembering that made me reconsider.

The Morris Worm: A Real-World Preview

The Morris Worm is the clearest real-world preview of how AI could go wrong — not through malice, but through a gap between what a designer imagines and what a system actually does once released.

In November 1988, Robert Tappan Morris, a Cornell graduate student, launched a program meant to measure the size of the internet. A bug in the replication code made it spread far faster than intended, and it took down roughly ten percent of the network — about six thousand machines, a huge share at the time.

Morris became the first person convicted under the Computer Fraud and Abuse Act.

Disasters Don’t Always Start with Villains

The personal irony is what makes the story stick.

Morris’s father, Robert Morris Sr., was chief scientist at the NSA’s National Computer Security Center — the office responsible for protecting the very networks his son disrupted. Contemporary reporting described him as “a father torn.” He had publicly condemned virus creators, compared them to joy-riding teenagers, and told the Times that nightly news coverage of his son was “not a career plus.” He met with the FBI, advised his son to stay quiet, and never expressed pride in the act.

That’s the shape of most real disasters. Not a villain with a plan, but a capability that outruns the designer’s understanding. The internet survived the worm — it got patched, got smarter, moved on.

What This Means for AI

Which is where the hype around AI deserves a second look. Most of the apocalyptic framing imagines something like a movie villain: an AI that “decides” to turn on us. That version is mostly noise.

The Morris Worm points to the real mechanism, and it’s a lot less dramatic and a lot more plausible — a system released with good intentions that behaves in ways its own creator didn’t foresee, at a scale no one had time to stop.

That’s not a hypothetical about some future superintelligence. It’s just software, and it already happened once.

The Real Question

So there’s an end to the truth in the doom talk, but not a zero. The question worth asking isn’t “will AI wake up and decide to hurt us” — it’s whether we can keep closing the gap between what these systems are built to do and what they actually do, the way the internet did after 1988.

The difference this time is speed and reach. A grad student’s bug took down six thousand machines.

The next gap might not give us that much warning.

Have ideas, questions, or thoughts to share? I’d love to hear from you—feel free to reach out anytime at g.*******@***et.com. And if you want more insights like this, subscribe on our website to get regular tips on cybersecurity, IT management, AI tools,  and more delivered to your inbox weekly.