Beyond man vs machine: Human and AI collaboration is the future of OT cybersecurity

By Richard Farrell

FROM the electrical grid, transportation, power plants, manufacturing of goods, down to a city’s water and sewer systems, operational technology (OT) systems control the critical infrastructure that we depend on.

Yet, OT cyber risk remains one of the most pressing concerns for organisations around the world today. Close to 80 per cent of OT organisations experienced more than three intrusions in the past year. For Singapore, nearly half of the companies that suffered cyberattacks were shut down last year, emphasising the dangerous and disruptive nature of these OT cybersecurity breaches.

Still, preventing these breaches isn’t straightforward. The expanding cybersecurity skills gap of 3.4 million people worldwide is a growing concern, with security experts citing the lack of training needed to cultivate OT experts as a root issue that needs to be addressed.

That’s not all. Organisations are also beginning to realise that simply duplicating cybersecurity strategies from information technology (IT) to OT proves impractical – and certainly unsustainable. Unlike IT systems, OT devices tend to be purpose-built, running on proprietary software and systems that differ from the norm. They are also often installed in remote locations or harsh environments, making access to them difficult. As a result, these systems aren’t updated as often as IT devices, and may have numerous software vulnerabilities.

The rise of artificial intelligence (AI), however, presents an opportunity to close this cybersecurity gap.

As AI continues to mature and become increasingly accessible, new possibilities around AI applications are also emerging. The booming success of ChatGPT-4, for instance, has propelled the use of AI to the mainstream. From lightening the workload of students to planning travel itineraries, many have scrambled to the portal to experience the quick and convenient service of this virtual assistant, while tech giants such as Alibaba and Google roll out rivalling services in the hopes of matching up.

Today, many organisations are positively considering the implementation of such technologies in OT systems to automate specific cybersecurity functions. In light of existing cybersecurity skills gaps, organisations are also looking towards AI to make up for the shortage of experienced OT professionals to properly secure their systems and assets. According to IBM, 64 per cent of OT executives have already implemented some form of AI in their security systems, and only 7 per cent of business leaders are not looking into automating their OT capabilities.

The benefits seem clear. At the minimum, AI takes away the need for full human control, reducing the possibility of human error and freeing up humans’ bandwidth for other essential tasks. It can also quickly and efficiently analyse a situation and make the best decision in a critical situation.

Collaboration, not replacement

The deployment of AI-based tools, however, is not a silver bullet to tackle cybersecurity needs.

Many OT systems are highly complex and consist of segregated technology. Given the critical nature of infrastructure that employ OT systems, human intervention and inputs are often needed to ensure that the automated technology can interact and integrate successfully into current security measures. OT experts must also have a thorough understanding of the technologies they use, as even the smallest lapse can have numerous cascading effects.

To better protect OT systems and reduce the stress human experts face in their roles, AI should be seen as an enabler of improved efficiencies, rather than as a complete replacement of preceding technologies.

This can be done by using a layered-defence approach. Automation can come into play by providing overlapping protection for threat detection and response, and monitoring for cyberattacks during seasonal or peak periods of high usage. Specialist AI technologies can also aid experts by parsing OT systems data and detecting patterns that help identify cyberattacks. With AI aiding the design and implementation of multiple layers of security around a product or system, OT cybersecurity experts will be able to focus on more crucial layers of defence within the system, and human errors can even be reduced.

Beyond implementation, organisations should also prioritise maintenance and upgrading of automated cybersecurity technologies rather than constantly opting for new software. This includes conducting penetration testing, vulnerability assessments, security audits to identify any weaknesses or gaps in the organisation’s system as well as the software used to defend the system. Updating software on a timely basis also helps to patch security flaws, ensuring that the organisation remains secure in the face of attacks.

With the AI-based cybersecurity market projected to grow to US$46.3 billion in 2027, it is clear that automation is here to stay and will only continue to grow more commonplace within the sector.

While AI has great potential in OT security, its intelligence is limited to the data it receives, and cannot override the need for human inputs in the workspace. OT experts must take a holistic approach in securing their products and systems, allowing AI and automation technologies to assist with tasks, while simultaneously forging the human-machine alliance that will spur innovation and propel businesses forward.

The writer is Asia-Pacific regional director of cybersecurity, digitalisation and data centre businesses at Eaton.