KorvaJoin the waitlist
Tech

October Brought An Unprecedented Wave Of AI-Powered Attacks

Wiz's technology chief says AI can now find and exploit complex flaws, and open source models are spreading the skill.

AI image
The a16z Show · Every Day Is Zero Day | Wiz CTO Ami Luttwak

In October, attackers using artificial intelligence found and exploited serious software flaws at a speed and scale never seen before, in what Ami Luttwak, the chief technology officer of cloud security company Wiz, calls Black October.

The shift matters because AI can now do work that once required elite human hackers, putting companies, startups, and even connected household devices at risk at the same time.

Luttwak said October brought both an unprecedented number of vulnerabilities and unusual severity, including a flaw in PHP, widely used software for running websites, that he called "probably the most severe vulnerability ever found by AI."

He pointed to two earlier warnings. One was what he called the Mythos moment, when an AI system showed it could find vulnerabilities but little happened afterward. The other was an incident involving OpenAI that he described as an accidental AI attack.

In October, he said, those warnings became real. Models are now good enough to find and exploit complex flaws in practice, and open source models are catching up much faster than expected, spreading the capability beyond a few leading labs.

The result is daily zero day attacks, or attacks that exploit previously unknown flaws. As an example, he pointed to a recent attack on a kernel driver, the core software that lets a computer talk to its hardware.

Asked about an April post that warned of a Y2K-like moment in 12 to 18 months, Luttwak said Black October may be its beginning. He said companies with systems exposed to the internet will need new security designs or should take those systems offline.

Questions people ask

What is a zero day attack?

A zero day is an attack that uses a flaw the software maker does not yet know about, so there are zero days of warning before it is used. It is dangerous because there is no patch available when the attack starts, leaving defenders to detect the intrusion and limit damage while a fix is built.

Why does a flaw in PHP matter so much?

PHP is widely used software for building and running websites, so a serious flaw in it can affect a large share of the internet at once. The risk is greater because some systems cannot be patched quickly, which can leave them exposed for months or years.

What is an open source AI model, and why does it matter if it catches up?

An open source model is an AI system whose underlying code and weights are publicly shared, so anyone can download and run it. When such models gain hacking skills, the ability to find flaws spreads from a few advanced labs to many more attackers.

More in Tech