CIN’s AI for Network Security System project unveils CASPER AI at UTS Cybersecurity Precinct
On Friday 21 November, the Connectivity Innovation Network’s AI for Network Security System project reached its final milestone with a live demonstration of CASPER AI at the University of Technology Sydney’s Cybersecurity Precinct.
Developed by a UTS-led team for the Connectivity Innovation Network, CASPER AI (Context-Aware Security Policy Enforcement and Response) is a virtual assistant designed to help detect, interpret and respond to cyber threats by drawing on information from across an organisation’s systems. The project was delivered by the Call for Proposals winners, Dr Xu Wang and Professor Ren Ping Liu, from the University of Technology Sydney.
CIN Director Dr Ian Oppermann opened the event with a throwback to CIN’s Cybersecurity Symposium in June 2023 where we heard from cybersecurity NSW Government, Industry including AWS Partners and Microsoft and Telcos. The Symposium and workshop identified many shared cybersecurity challenges across all sectors. Some of the insights that stood out were the fact that approximately 75 percent of cybercrime is identity related and there is a lack of general cybersecurity awareness and engagement in the workforce.
To address some of the shared challenges that were raised at the symposium, several recommendations were made:
- Basic cyber hygiene could prevent up to ninety eight percent of incidents
- Cybersecurity should be a shared responsibility across organisations, not just an IT function
- AI-driven threats such as deepfakes are becoming more sophisticated, requiring more advanced, behaviour-based detection methods.
The CIN’s AI for Network Security System project resulted from the Cybersecurity symposium insights and recommendations, thanks to our industry members, guest speakers and government stakeholders
Unlike traditional security XDR (Extended Detection and Response) and SIEM (Security Information and Event Management) tools, which operate in silos or according to pre-defined or hard coded rules, the novel AI security assistant developed by UTS leverages the power of large language models (LLMs) and multimodal reasoning to support flexible , context-aware security support. Dr Xu Wang demonstrated how CASPER AI can:
- Analyse information across multiple systems, including security logs, HR, travel and expense and physical access systems
- Build a richer picture of what is happening across an organisation by understanding context
- Detect anomalies that may not be visible from a single system or even a connected ecosystem of APIs and platforms
- Help security teams take policy-aligned actions more quickly through automated workflows
- Protect people and communities, especially those with lower levels of digital and connectivity ability, from cybersecurity threats such as fake access points.
Dr Xu Wang ran three demonstrations of CASPER AI in real-world scenarios. The first followed the deployment of a Cell On Wheels (COW) after flood eventing in Wagga Wagga. CASPER AI to pull together policies, equipment specifications and local context to support fast, secure planning.
The second demonstration detected a physical imposter, where discrepancies between a COW engineer’s approved travel, remote login and physical access records indicated “impossible travel” and identity possible identity misuse, illustrating how CASPER AI can also surface insider threats and unauthorised equipment activity. In this case, upon flagging the anomaly, CASPER AI can find and review vast amounts of information to recommend and execute next steps in line with an organisation’s policies. This includes automatically drafting and submitting a security ticket via an existing web portal. This approach has the potential to reduce the time spent on coordination across teams and systems, freeing staff to focus on higher-value work.
The third demonstration showed how CASPER AI can prevent cybersecurity incidents, especially for people with lower levels of digital and connectivity literacy, who are at greater risk. In this scenario, CASPER AI is able to verify public Wi-Fi access points, including temporary networks established during emergency deployments. By automatically analysing access point details and comparing them against known configurations, CASPER AI can flag suspicious access points that mimic familiar network names, supporting safer connectivity for staff and communities with varying levels of digital literacy.
What’s next
With the AI for Network Security System project now complete, CIN will work with NSW Telco Authority and interested partners to explore potential trial settings for CASPER AI, from Public Safety Network use cases to broader applications across emergency management and critical infrastructure. Over time, ongoing research will focus on scaling CASPER AI’s contextual intelligence, automation, and integration capabilities to support cyber-resilient communities across NSW.
The project adds a new element to CIN’s portfolio of practical, place-based innovation projects focused on strengthening connectivity, security and resilience for communities across New South Wales.
