WIRED: AI-driven Flaws Pass 100,000 in 2026

AI-driven vulnerability discovery has pushed the annual tally of recorded software flaws past 100,000 in 2026, more than double the combined totals of the two previous years.

The surge coincides with the mainstream arrival of large language models. In 2022, the last year before LLMs went widely available, the National Vulnerability Database logged roughly 25,000 vulnerabilities. This year, with more than a month remaining, the count has already cleared 100,000, according to figures reported by WIRED. October alone brought more than 17,000 new CVE entries, and November has added over 10,000 so far, CNET notes — compared with fewer than 5,000 in October 2025.

Volume up, signal steady

Despite the explosion in raw numbers, the pool of vulnerabilities that actually matter to defenders has not grown at the same pace. “The number of vulnerabilities has certainly exploded, but the number of vulnerabilities that you need to care about has remained somewhat constant,” said Slowik, cited by WIRED. “The caveat is that the pace at which they’re being used or weaponized has increased dramatically.”

Kernel panic brace for more outages security
Photo: WIRED

Patching can’t keep up

Finding flaws is only half the problem. “AI can help, but people still need to understand what went wrong, check that the patch addresses the cause, and look for similar issues elsewhere,” said Lea Schönherr, a tenure-track faculty member at CISPA Helmholtz Center for Information Security, quoted in WIRED. “Rolling out patches across widely used systems also takes time.”

Two-year outlook

Matthew Prince predicts “it’s going to feel like the sky is falling for two years.” But he expects AI to eventually help defenders gain the upper hand on common vulnerabilities. “These new threats and new security vulnerabilities won’t happen as often, and the background noise will actually tick down quite a bit,” he told WIRED.

A digital background in binary with a crack in it
Photo: CNET

I think software is going to be an order of magnitude more secure by the end of the next year or two than it is today for this reason. It will hurt a bit. John

The Washington Post framed the moment as the start of an “age of endless digital breakdowns,” noting that when the National Security Agency’s classified messaging system went out of commission Tuesday, speculation immediately turned to foreign adversaries, cyberattacks, or AI agents gone rogue.

“You’ve got to patch. You go for a flu shot, and you go and have your preventative healthcare,” said Adam Meyers, speaking to CNET. “You need to do the same thing when it comes to cybersecurity.”