Charting the evolving role of AI in facilities cybersecurity

There are two sides of security when we talk about artificial intelligence; how the technology can protect facilities operations, and how to safeguard the AI itself.

By Cijo Joseph and John Cruise

When we’re developing new services, Mitie is always ‘customer zero’. Everything we deliver for our clients starts as a way for our own teams to explore new opportunities, work more efficiently and manage our own systems. This means all our technologies and processes are tried, tested and refined before they become part of your facilities management package, including artificial intelligence services.

Our role as customer zero is especially important when we’re talking about security. When you’re introducing a new technology into your facilities management strategy, you need to be confident that it’ll perform well without putting important systems and data at risk; that’s why we’re committed to being transparent in our approach about everything from ethics to cybersecurity.

Within Mitie, we use AI extensively within our Security Operations Centre for predictive analysis, and to make sure our systems are constantly updated with the latest threat intelligence. So, we know how instrumental AI can be for safeguarding estates, but we also understand that it’s vital to keep the technology and data itself secure too.

Protecting your estate, armed with artificial intelligence security

The digital transformation of facilities means more oversight and insight for managers, but it also means there’s a growing number of devices, sensors and connected assets to monitor and protect. In Mitie we combine IT (Information Technology) and OT (Operational Technology) through IoT (Internet of Things). All these different inputs, combined with various security tools, can lead to a complex security landscape – especially if the estate is spread across multiple sites or facility types.

AI can simplify a lot of this complexity, by streamlining data consolidation and handling some of the time-consuming tasks that can draw security teams away from high-value work. By automating workflows, predicting outcomes through AI and Machine Learning, and removing the need for hands-on monitoring, those teams have the time and information needed to fully investigate incidents and use their experience to make the right call.

Analysts aren’t superhuman – AI can take care of the data overload

An effective set of security measures will constantly log new information about user behaviour, potentially suspicious activity and application security. These large volumes of log data are far too much information for a security analyst to parse quickly and effectively. But AI can correlate events to identify patterns – from unexpected logins to likely bot activity – and highlight information that could be overlooked by the human eye.

We talk about predictive analysis frequently – usually because of its role in maintenance and asset management. But it also has a crucial role to play in understanding security threats. AI can learn rapidly, understanding what normal operations look like and recognising deviations from that expected activity. With that knowledge, it can then spot early indicators of potential cybersecurity attacks and prevent hackers or malicious actors reaching important systems or information.

Generative AI (GenAI), the newest iteration of AI, can also help security teams make the most of the threat intelligence data from your estate and additional sources. GenAI can use natural language processing and generative capabilities to summarise large volumes of intelligence, identifying patterns and red flags that can be assessed by analysts or automatically uploaded to security tools.

A secure estate demands secure tools and technologies

In choosing how we use AI within Mitie, we’ve taken our own advice about getting the right experts for the task. So we’re building on the solid foundations of large language models and other AI tools from Microsoft, IBM, AWS and various industry leaders. Each of these partners has its own strong security measures and credentials, which we augment with our own facilities-specific frameworks.

Our partners take care of some of the fundamental security needs of AI, such as ensuring the data used to train the model is free of malicious content and hasn’t been ‘poisoned’ to compromise security. Once the model is in use by our teams, we have stringent protocols in place that make sure we’re protecting what’s important, including:

• A ‘defence-in-depth’ approach that uses multiple tools to safeguard the network’s perimeter, applications and end-user devices.
• Strict permissions that ensure sensitive data can only be accessed by specific individuals.
• A zero-trust network design where virtual desktops limit direct access to critical applications.
• Firewalls, intrusion detection and other measures to protect internal and external interfaces.

The future of facilities security is still in human hands but fully enabled and supported by AI technologies

Security threats are changing all the time – and that means security measures need to exceed their pace of evolution to make sure we can safeguard our sites, systems, data and people.

We’re dedicated not only to driving this evolution forward, but also to maintaining the right balance of technology and human expertise. Applied in the right ways, AI can be exceptionally useful in protecting systems and users from malicious activity – but it still needs human oversight to verify its conclusions, guide its decision-making and have the final say.

Learn more about our approach to artificial intelligence projects, and the standards we’re setting for ourselves and the services we deliver, and how Mitie is redefining the future of FM with AI.