One in Four Breaches Is Now AI-Enabled. Here's Every Way That's Shown Up in This Series.
New Research Synthesis August 2026 · 8 min read For most of this year, the AI-powered attacks we've covered read like isolated, almost freakish stories — a subscription phishing kit, a ransomware attack that fixed its own error in 31 seconds, an email assistant that helped rob its own employer. IBM's newest global breach research shows they were never isolated. One in four malicious breaches now involve AI, up 56% in a single year, and they cost businesses a million dollars more than an ordinary breach. Here's the number, and every place it's already shown up in stories we've told this year. Every time this series has covered an AI-related attack, there's been an implicit question sitting underneath it: is this a genuine trend, or a handful of striking anecdotes that make for a good headline? A subscription phishing kit is alarming. So is a ransomware attack that fixes its own mistakes. But two or three ...