Is it time to trust AI powered security?
19 January 2018
- By Paul Griffiths, Senior Director, Advanced Technology Group, Riverbed Technology
Last year, Google DeepMind’s AI program AlphaZero took 4 hours to master the game of chess, exceeding the realms of human capability. This is just one of the many existing examples of AI’s dramatic acceleration in the past few years. When we think about this from a cybersecurity perspective, the perpetual game of high stakes chess, many businesses are realising they are much stronger with AI on their side. Despite this, there is a high level of mistrust in AI that exists throughout the corporate landscape.
This year is set to be another year of learning for many businesses. The General Data Protection Regulation (GDPR) is right around the corner, and enterprises are beginning to understand that their current security ecosystem will soon no longer be enough to protect them in the long term. With a constantly evolving threat landscape, it is vital that investments are made into technology that will enable them to stay agile and outpace any threats.
Handling an organisation's security utilising a static platform is already an outdated practice in the cybersecurity space, and human error is far more often the weak link in the network security chain. Artificial Intelligence (AI) and Machine Learning (ML) is increasingly built into the best-in-class cyber security products, despite the fact that businesses are perpetually sceptical of enabling AI to have a powerful role in such a vital area.
The Immune Response
Whenever you’re feeling a bit under the weather, no one can know you’re beginning to get ill than yourself. Similarly, when it comes to monitoring the network, nothing can learn what ‘normal’ looks like, or respond to any drastic changes, more effectively than monitoring technology that sits within the network itself. If CIOs and IT departments rely on staff to constantly monitor their network, in order to identify and act upon incoming threats, they will inevitably lag behind in the cybersecurity stakes. Therefore, it is vital that businesses recognise the value of network performance monitoring tools, sooner rather than later when it comes to integrating security throughout their infrastructure.
READ MORE: Six ways of achieving endpoint security
The advantages of building your own AI “immune systems” are countless. Not only do businesses benefit from the way solutions can adapt to the threat landscape as it evolves, but it also eliminates to a degree, the issue of human error that plagues enterprise security. The security benefits aside, enabling a network to protect itself will reduce the workload for IT staff, enabling them to add value to other areas of the business, and scale up and down when needed.
With any AI based solution comes fear and a lack of trust, whether this is in regards to self-driving cars, or the tech inside your phone. However, while a cautious approach is necessary with any breaking new technology, rejection and refusal to adapt will leave you behind your competitors very quickly. There are a number of key barriers that must be addressed in order to build momentum behind these solutions, both legal and psychological.
In the first instance, the required legal frameworks must be put in place so that companies can demonstrate they are in compliance with the appropriate standard. GDPR is already shaking up the security world from a legislative standpoint, so this must be considered while the sector is under review to ensure consistency.
There is also a very human-led reluctance to embrace AI in the workplace. The feeling of being replaced and made redundant is one that is rippling throughout all sectors of work. However, when AI is utilised in the right way, it is ultimately a tool to make jobs easier rather than something that will take over the job role completely. Look, for example, at the driverless car debate. Many road users are reluctant to take their hands off the wheel and let cars drive themselves, despite the fact this is what airline pilots have been doing for decades with great success.
AI and ML are slowly but surely transitioning into both minor and major aspects of our home and work lives. Therefore, when handling something as vital and high stakes as enterprise security, it is only natural for the transition to be met with resistance. However, it is vital this transition is made, if businesses are to keep up with the rapidly evolving threat landscape.
As with driverless cars in the near future, security teams will need to learn to take their hands off the wheel and give more control to the technology that sits within the network they are trying to protect. Failure to adapt may well be costly.