Something is very wrong with our approach to cybersecurity. We are not winning the war. The bad guys keep getting through with more and worse attacks every year. Why is this happening? For one thing, the war is asymmetric. Their malware takes just a small amount of code, but our defense systems have grown to be millions of lines of code. Additionally, since the processors in use today are simple, they can’t tell if they are under attack. And to exacerbate that, all software code is buggy, providing an open door to sophisticated attackers.

There is an unrealistic fear that AI represents a threat to humanity because of the scare created in 1991 by SkyNet in the movie Terminator 2. But there is a real and present danger for AI and us from cyberattacks. As the National Institute of Standards and Technology puts it, “These technologies are also vulnerable to attacks that can cause spectacular failures with dire consequences.”

Cybersecurity discussions, whether in blogs, conferences, forums or elsewhere, tend to focus on concepts like anomaly detection, use of AI in cyber defense, security hygiene, phishing or ransomware. The cybersecurity technology known as compartmentalization has never come up when I have been within earshot.

To make progress, we as a society are going to have to address cybersecurity in hardware. But it is possible to write better software, and for that, we need to train students and employees in the most modern programming languages, such as Rust, which do a better job of avoiding common bugs.