First of all, unless it is a really bad AI, the answer is no. Now that we got that out of the way, here is why that matters.
Up until now, cyber defenders have been fighting against other humans. While the speed of the hack/defend cycle is not perfect. hackers and defenders are somewhat equal.
Think of a sports metaphor.
Pick any two NFL football teams. While on any day one might be better than the other, it usually is not a blowout. And even when it is a blowout, it might be 60-3. I don’t know of any pro football games where the score was 235-3.
But now consider a human defender against an AI bot. The bot can think faster, type faster, counter your defenses faster. In fact, it can do everything faster.
That includes royally screwing up.
You are already fighting against AIs and it is only going to get worse. Much worse.
So, if you think you can defend against AI with humans, you have not been listening.
Therefore, assume you are going to use AIs to counter any attack. Most of the time that will work. Maybe better in some cases than others.
But, you are going to have to let go of the leash at some point and that is when you have to worry about your attack dog going after the wrong thing.
In a sense that is what happened to CrowdStrike, even though that wasn’t a cyberattack – or at least no one is saying that it was one. They thought their system was perfect and everyone saw what happened. It was not pretty.
But at some point, like it or not, you are going to have to let go of that leash.
We have already seen that AIs can beat any human chess grand master. In that case the consequences are limited to bruised egos.
If an AI-driven attacker succeeds, the consequences could be much more significant.
On the other hand, What if an AI-based defender goes off the rails? Remember when Google’s AI responded to the question about how to keep veggies on your pizza? It suggested gluing them on. Probably not optimal. The thing about that answer was that a human was able to review it before getting the super glue out. If a fully automated defender goes off the rails then the results could be worse than the attack because inherently, the AI defender will have a lot more permissions at its disposal.
While there is no easy answer, now is a good time to start thinking about it.
Questions? Contact us. Credit: Data Breach Today