CyberSecurity SEE

Where Are Your AI Models Tonight at 10 p.m.?

Where Are Your AI Models Tonight at 10 p.m.?

potentially harmful, you have to quarantine it, you have to contain it, and then you have to neutralize it. And those three components are not easily done today.”To get ahead of the problem, Dehghanpisheh recommends bringing in security professionals to work side-by-side with data scientists, ML engineers, and developers, much in the way that DevOps has integrated security into development and operations teams in traditional applications and cloud systems.”I think doing the same thing for your AI group is very, very important,” he says. “It’s all about shifting left — that same term that we keep throwing around in the security circles. And that is super important because you can’t bolt on security after you build it; you’ve got to build it in.”Ultimately, he knows that this is going to be a long and hard fight to get AI supply chains under control. But it’s one that must be done, he says, given the increased likelihood that AI systems will be coming into automated cyber conflict scenarios with adversaries who are going to be using AI themselves. This means there’s a pretty good chance that if you’re not engaging your AI defensively today, you may find out the hard way that it’s getting engaged offensively tomorrow.”If you don’t start protecting your AI today, you’re going to be in a world of hurt tomorrow,” he says. “The early businesses are going to be the winners in the AI age. And unfortunately, today’s adversaries are just as happy to win as well.”One thing’s for certain: the AI security challenge is looming large in the distance, and it’s not going away anytime soon. In fact, it’s only going to get harder over time. With the explosion in AI models being created and shared on a daily basis, companies will need to get serious about understanding the unique security implications of these models or face the potentially disastrous consequences. It’s a tall order, but in this age of accelerated digital transformation and AI innovation, it’s a critical one to tackle head-on.

Source link

Exit mobile version