Matthew Rosenquist explains how attackers are using AI to boost social engineering, with 70-80% of these attacks aided by AI tools as of late 2025. So far the advantage is small: translated messages and fake images. But defenders lag behind because attackers move fast while defenders need reliable systems, giving attackers a dangerous head start.
Chapters
We see attackers using AI to evolve their attack, and we're already seeing it and feeling it. Even by the end of 2025, there were estimates between 70% and 80% of all social engineering attacks were being benefited at some level by AI tools. Now, to be honest, it benefit, but it was pretty small. In most cases, it was translation to local language. It was creating an AI image or something. Small, that will increase. So we see the victims, we see the industry embracing AI as fast as humanly possible. And again, that brings with it potential vulnerabilities. On the defender side, anytime the attackers get too big of an advantage, we have to respond and we have to get systematically better. And that's what's happening. We're going to see AI coming on the defense side. And at the end of the day, certain things hold true. Attackers always maintain the initiative, so they get the first mover advantage and they get to use the tools in haphazard ways, whereas the defenders really don't. So they have a time advantage, a window between when they get to use the tools and when the defenders actually get their mature tools.
