AI Will Soon Be So Good At Hacking, We¡¯ll Only Be Able To Stop Them With Other AI

As AI research progresses further and further by the day, we¡¯ve come to a point where we have to consider how the technology can be misused as well, if we¡¯re going to build precautionary measures. And while some experts believe our greatest threat from AI is in their use for military purposes or stealing our jobs, they¡¯re just as dangerous when being controlled by malicious humans.

Even, as early as this new year, we may see autonomous systems that are capable of hacking into our websites and networks and carry out all kinds of cyber attacks. AI after all can work independently towards a goal without needing hand-holding, can make decisions and adapt on the fly, and can outperform us when it comes to calculations and data analysis. That makes them perfectly suited hacking tools for the new age.

We may soon see machine learning algorithms repurposed towards hiding a piece of malware on a network, using its pattern analysis to ensure that the malicious code¡¯s actions blend in with the daily traffic. Say a company manages to get a bit of spyware into a major rival¡¯s network. It can then stay dormant, while the AI analyses data over time to learn crucial details ¨C how does the firewall work, what time of the day are cyber security teams active, etc. The AI can then use this data to find a weak point in the security system, and even create cover for the spyware to siphon and transfer out data on new technology being researched, or major projects being negotiated, or even which employees are unhappy and looking to quit.

And that¡¯s just in corporate espionage. There¡¯s no telling how much havoc an AI-powered malware can wreak on regular consumer systems across the Web. DDoSing will become so much easier when artificial intelligence can pinpoint and take over improperly secured IoT devices and more.

More worrisome, we already know how good Ai has become at manipulating video and audio of real people. Aside from that, it¡¯s only a matter of time before their analysis can let them impersonate how someone writes emails or posts on social media. How will you even know if the email from your boss was really penned by him or was instead created by a string of ones and zeroes?

And these AI will be too smart and too subtle to fight as we do now. Instead, we¡¯re going to have to take a step back, and let other AI deal with these troublemakers. And that just brings in another whole world of doubt and apprehension.