We all know that antivirus programs, while certainly essential, have their limitations, but an Israeli security startup is working on algorithms which could give security suites even tighter defences in the future.
This all revolves around virus definitions, the updates that security companies regularly deliver to their products – some being more regular than others, but there's always something of a delay in malware hitting the wild, and antivirus software receiving the appropriate updates to detect it.
As MIT Technology Review reports, Tel Aviv-based Deep Instinct (yes, it sounds more like a film than a security firm) is busy working on proprietary deep learning algorithms – AI neural networks which continually learn like the human brain, and can recognise fresh malware as it emerges with no need for definition updates to be piped over.
Or that's certainly the grand theory, anyway, although any improvement in early recognition of new types of malicious code will certainly be most welcome.
This sort of system will be particularly helpful when it comes to existing strains of malware which have only been slightly modified and tweaked, in an attempt to evade detection by your typical antivirus program.
Dr. Eli David, head of Deep Instinct's deep learning research group, and co-founder and CTO of the company, said that in their own testing, the system had already proved it could detect 20% more new malware when compared to existing security software.
Of course, other efforts have also been made in the deep learning malware foiling sphere, notably by a group of researchers from Microsoft. (Deep learning is a big thing over at Redmond, which has a Deep Learning Technology Center working across multiple initiatives including big data analytics, knowledge management and natural language processing).
It might not be too long, then, before we see the technology further honed, and being deployed to considerably bolster commercial antivirus software.