Palo Alto Networks Adds Deep Learning Algorithms to Security Platforms
Palo Alto Networks today released an update to its firewall’s core operating system, PAN-OS, that expands the depth of the artificial intelligence (AI) capabilities that can be used to thwart cybersecurity threats.
Code-named Nebula, version 10.2 of PAN-OS added deep learning algorithms that Palo Alto Networks can apply to, for example, instantly identifying zero-day threats or malware that is using specific techniques to evade legacy cybersecurity platforms. Previously, Palo Alto Networks was using machine learning algorithms to thwart cybersecurity attacks. Deep learning algorithms expand on machine learning algorithms’ capabilities using neural networks that are capable of learning from the security events they observe.
As a result, Palo Alto Networks claimed that the intrusion prevention capabilities it provides can block 48% more malware employing an unknown command-and-control system than any other solution. The company also claimed its AI-infused solution is now six times faster.
Anand Oswal, senior vice president for network security, said that capability means AI can now be applied inline to thwart attacks as they are being launched. Previously, security analytics was applied after the attack has already been launched. Much of that capability is enabled by optimizing AI performance in the core processors Palo Alto Networks uses in its firewalls, noted Oswal.
Other improvements include updates to advanced URL filtering capabilities that Palo Alto Networks said prevent 40% more threats from malicious websites, a 40% increase in the amount of DNS traffic that can be inspected and the ability to discover 90% of all devices within 48 hours of being connected to a network. Policies can now be applied faster, as well, the company claimed.
Finally, Palo Alto Networks now uses AI to analyze telemetry data to better predict firewall health, performance and capacity issues to prevent up to 51% of disruptions that previously might have impacted firewalls.
There’s no longer a question of if but rather when and to what degree most organizations will adopt some form of AI to combat cybersecurity threats. While algorithms are not likely to replace the need for cybersecurity professionals, it’s apparent given the increased volume and sophistication of the cyberattacks being launched that there is a growing need to augment cybersecurity professionals’ expertise with technology. Most organizations are finding both hard to find and retain.
Naturally, there is always a fair amount of skepticism when it comes to anything involving AI. However, as deep learning algorithms are more widely adopted, the amount of time required for AI-infused platforms to learn an IT environment should be reduced. Much like any new employee, it takes time for algorithms to create a baseline that enables them to understand what constitutes normal behavior in an IT environment. On the plus side, those algorithms never forget something once they learn it.
In the meantime, it’s not clear what impact AI is likely to have on demand for cybersecurity professionals. Today, there are millions of unfilled cybersecurity positions. Many of the rudimentary tasks that cybersecurity professionals manually performed are increasingly automated. The goal, however, is not to eliminate the need for cybersecurity professionals but to transform the necessary tasks to level the playing field against well-funded and well-resourced adversaries.
