Should Cybersecurity AI Models Be Open Source?
Dane Knecht says cybersecurity AI has to be open over time, because the people who need to defend will not all sit inside a few labs.
The same model can attack and defend. Safety guardrails do not split those uses into a clean pair, so an open defensive model is also an offensive one. Keeping the capability inside a few closed frontier models fails the other way. Those labs cannot hand it to every person who has to defend a system. Knecht's view is that the tools spread anyway. That is how open source tends to work. Attackers will get them, and so will the biggest companies. A closed release leaves out everyone else who needs the defense.
Democratizing Security for the Agentic Web with Cloudflare CTO Dane Knecht ·
From the transcript
the tool that's causing this kind of security apocalypse, lone apocalypse to happen. It is also the tool that can help us solve it faster as well. So it's on both sides of it. Where do you fall in this debate of the role, like good and bad, of open and closed models on the cyber front, right? Because the argument, the broad argument is, well, having defensive and offensive cyber capabilities, you know, they're parallel in the same model, despite...
uh, safety and guardrails being, being built into them. And so you can't have open defensive models without them also being offensive. And then the argument, um, sort of, uh, that were the concern around only having security capabilities in, um, a, a set of frontier closed models as well. How are you going to release them? And you can't possibly get them out to the distributed set of actors that need to defend.