Originally posted in IT Pro Portal on June 11, 2019.
Our lives online have changed at an ever-increasing pace over the past 15 years, with historic effects on the way we work, communicate and socialise. Streaming media, data analytics, cloud applications and mobility have transformed life and work profoundly, making possible what was unimaginable just a couple of decades ago. Digital transformation is making the data economy permanent and pervasive, and with the scale and value of information being put online, it’s increasingly imperative that the scale of defence be able to match it.
And inevitably, this need has completely outstripped skilled analysts’ abilities to keep up. Even using automated tools, cybersecurity operations centre (SOC) analysts are quickly inundated by alerts that can muddy the picture and increase uncertainty. While there are machine learning (ML) and artificial intelligence (AI) tools that can keep up with the new normal scale, chief information security officers (CISOs) are still struggling to balance where to place analysts and where to place machines in their cybersecurity strategies.
Malicious actors evolving at webscale
Given the vast quantity of valuable information available online – often with minimal or careless security, it’s no wonder that breaches, exploits and attacks have exploded. Recent attacks such as the Mirai botnet Distributed Denial of Service (DDoS) and WannaCry ransomware demonstrate the scope and breadth of cyber threat actors. Furthermore, 2.5 billion records were compromised worldwide in the first half of 2018 alone.
A crucial factor of this escalation of cyber threats is the availability of cheaper, easier access to technology with the democratisation of the tools and knowledge required to execute these sorts of operations. A broad offering of malware-as-a service options are available on the Dark Web and sold on a commission basis. Anyone who wants to make a fast buck and knows how to get on the Dark Web can become a hacker. The stakes have been raised.
Where human-scale falls short
Thoughtless mistakes, sloppiness and malicious intent, when combined with the complexity and scale of today’s digitised platforms, pose a serious challenge to traditional models of security that still heavily rely on security analysts not making mistakes. Many recent breaches are simply the result of human errors in complex environments.
There are a number of reasons why enterprise cybersecurity is increasingly complex and why securing it now requires an integrated AI and expert analyst approach:
- The newness of the technologies means security teams are unclear on how best to secure these applications. In June 2017, the names, addresses and account details of some 14 million Verizon customers were found in an unsecured data repository on a cloud server. This was not a result of a malicious attack; the repository was simply exposed to the internet because of an incorrect configuration.
- The attack surface is constantly growing, given the interconnectedness and cloud-hosting of many services. Most people assume their organisations use up to 40 cloud apps when, in reality, the number is generally closer to 1,000.
- Many organisations move to well-known cloud infrastructure technology companies because they assume those companies have better security practices. This is true in terms of the security of the infrastructure, but the cloud customer takes on many new responsibilities for configuring the available security settings and securing their own data.
- Many organisations move to well-known cloud infrastructure technology companies because they assume those companies have better security practices. This is true in terms of the security of the infrastructure, but the cloud customer takes on many new responsibilities for configuring the available security settings and securing their own data.
Our powers combined
It’s inevitable: machine-scale problems will demand machine-scale solutions, like machine learning. To successfully implement them, however, requires a conversation about how to apply these technologies in the right way, to augment the analyst, not replace them. The use of integrated machine learning can have a pertinent and powerful impact on its application in cybersecurity.
The cybersecurity paradigm needs to shift from finding low-level patterns in siloed data and then aggregating the output to instead looking for the patterns that matter in data that is aggregated across many sources if AI and ML is to be used effectively. Using these tools within an integrated approach will optimise the use of these new technologies, ensuring that ever-evolving threats can be detected and defended against. The promise of AI is to help organisations to automate the time-consuming process of analysing the data to understand a threat and to augment their analysts’ capabilities, who then must add context and determine how to react.
It is important to have an understanding of the three stages of change management in evolving from human-scale to machine-scale in cybersecurity defence. These are:
- People at the helm: Cybersecurity platforms based on AI should attempt to detect threats by monitoring instrumentation from multiple sources, and the results of the analysis should be delivered to a highly skilled analyst, who will then take action.
- Smart automation: Automating this process is the next phase change, to move away from the floods of alerts being generated by most tools today.
- Complete control: The use of integrated ML applied holistically across an entire enterprise enables AI-based cybersecurity platforms to achieve the best outcomes. These systems use automation to complete complex human tasks by using data from an entire system, not just a single focus point.
Looking ahead
The scope and scale of today’s cybersecurity threats is the new normal. Fortunately, the tools now available, such as artificial intelligence and machine learning, are increasingly able to meet the challenge. To successfully leverage new, advanced technologies to combat today’s ever-evolving threat landscape, human-machine interaction is what we need to work towards. Building integrated AI platforms that empower cybersecurity analysts is the new wave of change that the industry needs to make it effective against a rapidly evolving enemy.
