Next-gen antivirus systems will not rely on updates to detect all new threats


Cybersecurity solutions and their malicious counterparts are always at a constant battle and some may say that the good guys are not on the winning side this time. Malware is constantly created and distributed across the web and protection programs are constantly on the lookout for new threats with continuously updated databases of threat definitions. This creates an obvious flaw in the system where an antivirus application may be completely unable to protect against a certain threat if its database holds no information about it.

Researchers at an Israeli startup called Deep Instinct are attempting to remedy that situation by creating an antivirus application that can learn about threats and deal with them on the spot. Their software takes advantage of deep learning, a concept that allows software to process huge amounts of information and anticipate future events by “learning” about what is happening to the world around it and adapting to it.

Now that someone has done it, it seems quite obvious that deep learning could be used in antivirus applications. After all, there may be no other program type that receives as many updates as a protection app. The researchers claim that their service can spot new malware with 20% higher accuracy than the world’s current best antivirus applications and it will only get better in time. Other studies have supported this kind of approach to malware protection services so we may see them in the market soon enough, at least after the companies have mostly ironed out issues such as false positives.