DISPATCH

AI in the Shadows

How adversarial use of AI is changing the threat model. What we're seeing in the wild and what to prepare for next.

How adversarial use of AI is changing the threat model. What we're seeing in the wild and what to prepare for next. The shift AI is a capability, not a side. Adversaries are already using it for reconnaissance, phishing, and code. We're not in science-fiction territory-we're in the early, messy phase where tools are good enough to scale old attacks and occasionally enable new ones. What we're seeing Phishing and social engineering - More convincing copy, fewer telltale errors. Volume and quality both up. Reconnaissance - Automated scraping and pattern-finding. Faster mapping of attack surfaces. Code and tooling - Assistants used to write malware, refine exploits, and automate tasks. Skilled operators get more leverage. What to prepare for Assume that the cost of mounting convincing attacks will fall. Assume that defense will need to rely less on spotting obvious mistakes and more on behavior, context, and resilience. Now is the time to stress-test your assumptions about what "good enough" security looks like.