AI May Turn to the Dark Side in the Next Five Years

A recent report prepared by Finnish cybersecurity company WithSecure, the Finnish Transport and Communications Agency, and the Federal Emergency Management Agency analyzes current trends and developments in artificial intelligence (AI), cyberattacks, and the areas where these two fields intersect. According to the report’s authors, AI-driven attacks are currently very rare and are carried out in ways that make them difficult for researchers and analysts to observe.

Although neural networks are not yet capable of independently planning and launching cyberattacks, researchers believe that within the next five years, malicious actors will develop AI algorithms that can autonomously find vulnerabilities, plan and execute harmful campaigns, bypass security systems, and collect information from compromised devices.

WithSecure states that the first AI tools for cyberattacks will likely be used by state-sponsored hackers, and then will eventually fall into the hands of smaller groups, who will begin to use them on a larger scale. As a result, cybersecurity professionals need to start working now on systems designed to defend against such attacks.

The report’s authors also noted that AI-powered cyberattacks will be especially effective in the area of impersonation—a tactic most commonly used in phishing and vishing attacks.