Artificial Intelligence Is a Threat to Cybersecurity. It’s Also a Solution.

The last decade has demonstrated an unprecedented evolution of the ability of AI systems to learn and master skills that allow them to autonomously outperform human experts in a never-before-seen way. The autonomy given to these systems presents a very pressing cybersecurity challenge. These AI systems are trained using large datasets. Unfortunately, by tampering with the data used to train the AI, an adversary is able to manipulate how the AI makes decisions, and the absence of a human in the operation of the AI systems means that the compromise could go on undetected for a long time, leading to economic and life losses. The ability of modern AI, however, can also be leveraged to learn and detect the most subtle features of malware and other tools used for cyber attacks, which could help governments and corporations detect vulnerabilities in their systems before they escalate. Every institution needs to understand the risks and rewards and respond without delays.