Should you care about an “AI slowdown?”

Welcome to this week’s edition of the Threat Source newsletter. 

There’s been a lot of talk recently about slowing down the pace of AI development. And yes, there are legitimate moral, ethical, geopolitical, and safety concerns with the use of AI. It’s not clear yet whether an AI slowdown could happen, let alone whether it should (hat tip to Dr. Ian Malcolm). I admit, I’m not really qualified to opine on the impacts unrestricted AI might have on bioterrorism, the balance of international power, or even our chances of being eaten by dinosaurs. What I can tell you, though, is that any sort of “AI slowdown” is not likely to have much of an impact on cybersecurity.  

There are a few reasons to think this. The most obvious one is that models are already really good. We’re at the point where the newest models bring only incremental improvements in cybersecurity capabilities. Arguably, they’ve been getting better so fast that our ability to use them effectively for defensive tasks hasn’t kept up. On the offensive side, practically every recent model is already able to mine decades of tech debt to uncover an uncomfortable number of vulnerabilities. Instead of chasing model improvements, our best strategy might be to improve our agentic harnesses and frameworks, essentially giving us better capabilities with our existing models.