Artificial intelligence is clearly not a new idea. There are roots in antiquity dating far back beyond the science fiction works of Isaac Asimov. In a very real sense though, artificial intelligence has arrived. AI essentially involves building algorithms to spot bad behavior from good. It’s everywhere – facial recognition, emotional sensing algorithms, machine generated artwork, chatbots and more. AI is slowly but surely becoming prevalent in our daily lives.
The Good Side of Artificial Intelligence
Recently, IBM trained supercomputer Watson on was the ability to help clients respond to security threats. Within 15 minutes Watson can come up with a security analysis of a reported threat. For most humans, that specific task can take up to a week.
AI bots can save organizations time and money by going through structured data quickly and efficiently. Artificial intelligence bots can assist in 3 areas that are typically handled manually. First, is the identification of threats. Today our businesses have a large attack surface to defend. With applications, the cloud, mobile devices, smart watches, Bluetooth and other IoT devices, it’s hard to ensure they’re all protected. Second, is assessing and managing risks. Once threats are identified, bots look for any security gaps that could cause major issues if a breach were to occur. After identifying threats and the potential risks, the machine learning bots can then begin the remediation process. Machine learning allows humans to focus on threats that bots have not yet been taught to detect. There’s already a shortage of IT experts, so integrating AI into day-to-day operations is extremely helpful.
The Dark Side
However, there is a potential down fall when it comes to machine learning. Without proper supervision, the bots can lead humans to false alerts. Even more detrimental, it is increasingly difficult for designers to make machines that are not only safe but also capable. There are about 10,000 researchers working on artificial intelligence worldwide. Unfortunately, only 100 of them are fully focused on studying how to address the failure of an AI bot. Machine intelligence can go rouge for a multitude of reasons. This can include deliberate actions of unethical people or the side effects of poor design.
Cybercrime and AI
Another major concern about machine learning is that the bot will become a hacker. Cyber criminals are learning to manipulate machines with artificial intelligence. Hackers use AI bots to take results and allow other AI bots to “learn” what was found and then be used against other systems. An example of this is a bot finding a password and then passing along the information to another bot. Then the bots can exploit the system the password is being used for.
Cyber criminals also discovered ways to make AI bots to appear as if they’re the actual attackers. Machine learning bots can gather a lot of information quite rapidly. Sending out a tailored phishing email to those in the victim’s address book is a popular way machine learning is being manipulated by threat actors. With social media profiles becoming increasingly public there is more info than ever for a machine learning bot to sift through and learn, potentially leading to a potential phishing message or breach.
Is AI Worth it?
After learning more about artificial intelligence and cybersecurity, it’s understandable to have mixed feelings about it. AI is expected to fight off cybercrime and data breaches due to their high levels of intelligence and ability to act quickly. However, as we begin to obtain more access to machine learning, malicious actors will have access to it as well. The stakes may become higher due to better technology but it will also be easier to defend organizations with the help of AI.