Matthew RosenquistFounder & CISO Advisor, Cybersecurity Insights · CISO, Cybersecurity Strategist, & Industry Advisor

Defenders Must Prepare for Weaponized AI

19 Jun 2026 · 1 min

Watch on YouTube

0:56

In this brief monologue, the speaker urges security professionals to proactively engage executive leadership about the rapidly evolving risks posed by advanced AI models, especially as they become weaponizable. He highlights that Anthropic's Mythos is currently the most advanced model, with U.S. models trailing by only two to three weeks and Chinese models lagging by roughly two to two‑and‑a‑half months. He stresses that model performance is not plateauing but accelerating, creating a rising curve of capability that defenders must address as the new reality.

Chapters

Security professionals need to be working with our executive leadership and properly communicate the ramifications of these new models. And it's not just Anthropic's Mythos. All of the big AI companies are pushing their models to be able to achieve these kinds of capabilities and this speed and this scope as fast as they can. The US models are probably 2 or 3 weeks behind at this point, behind Mythos, but they're going to catch up.. And we're seeing overseas models, especially the ones in China, they may be 2, 2.5 months behind, but eventually they're going to get there. And Mythos itself is not staying stagnant. We've seen increases in the performance and new models coming out. So this isn't a plateau. This is a curve going up. This is the new reality that we have to deal with.