Here’s how DARPA trains machines to protect its assets

If the Defense Advanced Research Projects Agency (DARPA) were to get its way, security analysts and hackers would find themselves replaced by machines. In an ongoing project, DARPA plans to use artificial intelligence (AI) to tackle security issues and weed out any human competition in these areas.

On August 4, 2016, DARPA hosted the Cyber Grand Challenge at the Paris Hotel and Casino in Las Vegas, NV, where seven artificial intelligence algorithms squared off against each other to determine which was the best at patching a penetrated computer network. The event was live-streamed and opened to the public.

The results of past years have proven to be beneficial for DARPA. Much of the technology behind the self-driving car was a result of the Grand Challenge, as well as the robotic expertise likely to assist future missions on Mars.

There’s a logical reason DARPA wants to automate the business of cyber security. The time lag between detecting a network vulnerability and designing a solution for it gives hackers the advantage over security personnel. In the months between when a vulnerability is first detected and when a suitable patch is ready for release, hackers have free reign to move through numerous systems. Automating the patching process could help in favor of the security team.

But there is a challenge that comes with this goal of automating computer security. An artificial intelligence capable of writing computer code to patch a security leak could also be likely to penetrate a security leak. The possibility of an AI system capable of producing computer viruses more sophisticated than any human code exists. The idea of such a creation would be problematic and enough for foreign governments to believe America has an AI hacking agenda that could create a kind of AI arms race.

Currently, there are multiple governments using hacking tools to advance their political agenda. China has state-sponsored hackers and a virus called Stuxnet, which is famously used to undermine Iran’s uranium enrichment program. In light of this trend, a government-sponsored AI hacking program may be a primary motivator behind DARPA’s Cyber Grand Challenge.

The focus of the event centered around a game that hackers refer to as Capture the Flag, where the contestants tried to reverse engineer the components of a highly developed form of cyberwarfare to expose flaws and steal a specific file (the flag). At the same time, the contestants worked to patch security holes in their own system and protect the file which the opposition is attempting to command.

The results can be found here.

Source: ExtremeTech