AI at the frontline: How a new ProtectUK tool is transforming counter-terrorism support for public spaces
We interviewed Professor David McIlhatton, Director of the Protective Security Lab at Coventry University, about the new AI tool being developed to support how businesses engage with the advice and guidance on ProtectUK.
What is ProtectUK and who is it for?
ProtectUK is a platform that provides free and easy access to information, guidance and learning to help businesses and those responsible for venues and public spaces understand protective security and improve their response to the risk of terrorism. ProtectUK is operated by the National Counter Terrorism Security Office (NaCTSO) and supports the Protect and Prepare strands of the UK’s Counter Terrorism Strategy (CONTEST) to save lives and reduce the risk and impact of terrorism.
How did Coventry University become involved in developing this AI tool?
The Protective Security Lab at Coventry University have been working on advancing evidence and insights for protective security for quite a while. We work with problem owners in government, CT Policing and defence to understand the challenges that they have and we translate those into research and development questions. In this case, we identified that scaling guidance for the many businesses that would fall within (and outside) the scope of the Terrorism (Protection of Premises) Bill would be a challenge for a number of reasons, including the need to ensure consistency of message, the need for businesses to be able to understand the guidance, and for it to be accessible quickly.
What AI technology is being used?
The tool is designed to act like any other language model that people might interact with on a daily basis, such as ChatGPT or Copilot. But it is specific to the guidance that businesses would access if they were wanting to get authoritative information on how to protect their venues or public spaces from terrorism.
How will this AI tool improve on a traditional website?
We want businesses to have agency in how they manage their risks. Therefore, it won’t tell them what to do, but guide them on the types of things that they need to consider. We also want them to be able to access guidance quickly and not go down rabbit holes when looking for the answers that they need.
We think the best way to do this is in natural language, as people may ask the same questions in different ways. Others might not know what to look for when they go to a website and then be overwhelmed with all the information that might be on it. But we also wanted it to work offline in disconnected areas so that people had access to the information from anywhere.
Will users be able to generate policy and procedure documents for a specific site?
They will be able to ask questions and get answers that will guide and signpost the user to the location of the guidance. We don’t want the tool to create policy or procedure documents for businesses, and our research has demonstrated that this is the right approach.
Will this AI tool help understand trends or gaps for future guidance?
Absolutely. The research that we have undertaken to date has shown that there is a need to understand trends, particularly in the context of the questions being asked. These could be used to formulate things like FAQs that can then be shared out across larger groups or sectors – or be used to prioritise future guidance documents.
How are you ensuring this tool can be trusted?
That is what we are explicitly focused on in our research. Trust is critical from many perspectives and it’s why we are going through numerous research and development phases with businesses.
Can you give us some examples of using the tool?
The user can ask a question using speech or text for many different things. One example might be a user asking what might be needed in a site security plan, and another might be what to do if they have found a suspicious package. The tool will then respond based on that.
When will the tool be available and will be it free?
We want to ensure the tool is as good as it can be, but also that it meets the needs of both the business community and CT policing. That’s why we are in the process of carrying out extensive research with businesses on utility, usability, security, accessibility, messaging, as well as other things.
We want this to be underpinned by evidence and improved as we develop the tool based on the insights that we receive. Technology adoption usually fails when the users or potential users aren’t involved in its development. ProtectUK is free to access and open to all, so any new tools or features like this would naturally become part of an improved user experience.
