2025 Funded Projects

Project Description

Cyber-Physical Systems (CPS), Advanced/Urban/Electric Air Mobility (AAM/UAM/EAM), Autonomous Systems (AS),and Critical Infrastructure (CI) require practical man-machine interactions to ensure their survivability, resilience, trustworthiness, reliability, and performance objectives under malicious cyber-attacks. This projects develops novel solutions to secure AAM/UAM/EAM, AS, and CI systems. The solutions will ensure trustworthiness, safety, security, and management of these infrastructure. A key outcome will be an expansion of Canada’s expertise in this area, and it will enhance national security.​

The project will (a) provide solutions for AAM/UAM/EAM,AS, and CI to minimize impact of security threats, and maximize their cyber-attack detection, trustworthiness, safety, and resiliency, and (b) develop verified and validated solutions by utilizing high-fidelity hardware-in-the-loop (HIL) experimental environments.

Project Description

Addressing the critical cybersecurity challenge where disinformation spreads six times faster than factual content and digital attacks grow 15% annually, Koat.AI Inc. is developing a transformative defense solution through National Cyber Consortium funding. This groundbreaking 18-month initiative tackles the dangerous gap between when digital threats emerge and when organizations can effectively respond, replacing traditional reactive monitoring with autonomous AI agents that collaborate to identify, analyze, and respond to sophisticated attacks in real-time. By integrating cutting-edge deepfake detection capabilities with intelligent agentic AI workflows, the platform creates a proactive defense network operating at machine speed while maintaining human oversight, representing a paradigm shift from manual, fragmented security responses to an integrated, AI-assisted approach that anticipates and neutralizes threats before they achieve maximum impact. With over 70% of disinformation narratives remaining active for more than a year before platform intervention, this innovative solution’s continuous learning and adaptation capabilities ensure defensive systems evolve alongside emerging threat vectors, delivering the speed and precision required for today’s digital threat landscape.

Project Description

This project designs and develops a Digital Twin Platform to protect Canada’s critical energy-related infrastructure, such as power grids, transportation & communication networks, from cyberattacks. Testing & improving cybersecurity protections in real-world systems is difficult because it can cause service disruptions.​

This solution creates a virtual environment, or “Digital Twin,”that mirrors real infrastructure with realistic operating timescales. Using advanced technologies like real-time simulations, Artificial Intelligence (AI) & novel sequentialdecision-making solutions, this platform will allow utilities and industries to simulate cyber threats, assess risks, perform and practice predictive defense, solutions & responses without affecting actual services.​

This innovative platform helps Canadian organizations strengthen their defenses, comply with cybersecurity standards & frameworks, and ensure that critical services remain cyber-resilient. By providing a powerful research and innovation tool, the project will contribute to the safety & security of Canada’s critical infrastructure systems.​

Project Description

This project improves cybersecurity in the electronics supply chain by identifying and defending against hidden attacks introduced after devices are built. Today’s electronics contain hundreds of parts sourced from many suppliers. Attackers can take advantage of this complexity by secretly adding harmful changes to devices during shipping, installation, or maintenance. These changes can remain hidden and become active only under certain conditions, such as temperature or signal patterns.

The project addresses the gap formed by current security methods that often stop after devices are assembled and rely on trust through the supply chain. Existing tools to detect tampering often require breaking open the device or only work on known threats. This project explores new, materials-based attacks and studies how attackers can plan, embed, and scale them. It focuses on a defense technique based on radio frequency analysis to detect changes inside electronics without needing to disassemble.

The outcomes include a better understanding of hidden attacks, new detection tools using RF technology, and the training of specialists in both attacking and defending hardware systems.

Project Description

Despite the best efforts of technology leaders to train and raise awareness of important cybersecurity issues, employees continue to commit errors, accidents, and oversights that directly lead to costly and embarrassing cyber incidents. This project will address the managerial challenge of directing employee attention to the fulfillment of day-to-day cybersecurity responsibilities. Rooted in ahuman-centric view, we seek to understand how fluctuations in employee attention to cybersecurity responsibilities relate to behaviour. ​

The goal is to enhance employees focus on cybersecurity in day-to-day operations. They will develop and refine the collection of guidelines, frameworks, and tools that could be utilized by Canadian organizations to both troubleshoot problem areas, as well as offer meaningful solutions to employee attention to cybersecurity. This output will contribute to enhancing the secure behaviour of employees, thus reducing the potential for costly and inconvenient cybersecurity incidents.​

Project Description

This project develops a new paradigm in AI cybersecurity by enabling Large Language Models (LLMs) to defend themselves autonomously against evolving threats.​

LLMs are increasingly deployed in sensitiveand mission-critical environments such as healthcare, finance, and national security. However, current defense mechanisms rely heavily on static protections and manual interventions, which are often too slow and rigid to counter rapidly evolving cyber threats.​

The project will develop a self-healing AI security framework that enables LLMs to detect, respond to, and recover from cyberattacks independently. The concepts of LLM Auto Vaccination, which continuously exposes models to synthetic adversarial attacks, and Autonomous Patch Generation and Deployment, which empowers the system to automatically detect vulnerabilities, synthesize corrective actions, and deploy patches without relying on human developers are developed.

Project Description

This project will develop a smart, AI-powered platform that helps organizations understand and prepare for advanced cyber threats before they happen. Unlike traditional tools that rely on risk models, our system creates realistic “what-if” cyberattack scenarios using cutting-edge machine learning and simulation tools. It shows how threats like ransomware or zero-day exploits could spread across a company’s digital operations and what the impacts might be—such as lost data, system downtime, or financial loss. ​

By giving decision-makers a way to test defense strategies in a virtual environment, this project will help them respond faster and make better security investments.

Project Description

A controlled investigation into the utility of gamification to enhance cyber-safety for indigenous children. The study will determine if:​

• Kids get better at cybersecurity and achieve increased cyber-safety using a gamified approach over traditional pedagogical approaches. ​
• Language and stories are kept alive: The program helps pass down traditional knowledge and language in a modern format, which can make it easier for younger generations to stay engaged with their roots.​

The goal is to identify new ways to teach in the future. ​

Project Description

Agentic Artificial Intelligence (AI) systems are composed of autonomous agents that iteratively interact with dynamic environments, perceive feedback, and make decisions to accomplish complex, multi-step tasks, often beyond the capabilities of traditional AI models or large language models (LLMs). This new paradigm fosters an autonomous andgoal-oriented ecosystem. ​

However, the unprecedented autonomy and complexity of these systems introduce new security risks arising from unpredictable user interactions, intricate internal processes, dynamic operational environments, and interactions with untrusted external entities. This project develops a framework for systematic assessment and quality assurance of the security characteristics of agentic AI systems.​

We evaluate security aspects of agentic AI systems and mitigate underlying security threats in four ways: 1) cyber threat analysis and characterization, 2) security testing, 3)secure deployment, and 4) security monitoring.

Project Description

This project will identify and mitigate security vulnerabilities of AI agents. With a focus on untrusted, potentially malicious tools that can lead to unauthorized access to sensitive information or that bypass existing security protections. ​

A dataset of open-source AI agents will be curated, including sample user prompts, and examine each agent’s tools for security risks, documenting any harmful or unsafe tools with proof-of-concept prompts. Using this dataset, automated methods to test AI agents’ tools for vulnerabilities by modeling the AI agent as a state machine and using fuzzing techniques to uncover states where tools might behave maliciously or unsafely are developed.​

Simulated attacks that replace genuine tools with malicious ones in controlled experiments are developed to understand how such tools can compromise AI agents and whether they can evade detection. ​

Project Description

Artificial Intelligence (AI) models are becoming ubiquitous in applications such as healthcare, finance, robotics, fraud detection, and recommendation systems; but they require extensive training on private data, raising privacy and legal concerns. Fully-Homomorphic Encryption(FHE) enables secure, privacy-preservingcomputations on encrypted data that can be used in AI model training and inference. However, existing FHE frameworks incur prohibitive performance penalties, making real-world applications impractical. ​

This project advances privacy-preserving machine learning through accelerated homomorphic encryption and reduces processing time from hours to seconds, making interactive secure AI services viable for time-sensitive applications like medical diagnostics. This impacts use cases such as secure computations on private medical data, federated learning on edge devices, and protecting proprietary AI models. ​

The project designs a versatile, generalized hardware platform to support FHE computation using the novel approach of differential programming with practical processing-in-memory designs.

Project Description

This project develops a cybersecurity system to protect Canada’s aerospace and satellite technologies from cyber threats. Satellites, aircraft, and ground control systems are essential to national defense, communications, and navigation, but rely on older technologies that cannot be easily updated. New systems are not immune to hackers with more advanced attack methods, including those powered by artificial intelligence or quantum computing.​

This project will create a smart system that can automatically learn how aerospace systems communicate, detect unusual or suspicious activity, and explain the problem in clear, understandable language. This means even non-expert operators can understand threats and react quickly. A working prototype usable across multiple aerospace platforms, supported by Canadian defense and industry leaders will be deployed.​

Project Description

This is the first pan-Canadian project that addresses the pressing challenge of software supply chain security. First, we develop novel techniques and tools to secure the build infrastructure of software applications. These techniques will support developers in modeling and reasoning about the different automated steps that occur in their workflows. The goal is to prevent secret leakages and to develop new tools to support the rapid evolution and maintenance of the software supply chain. Every link in the software supply chain, from third-party dependencies to tools and workflows, releases new versions on a weekly basis. Our project will strengthen the supply chain with novel testing techniques to cope with these rapid changes.​​

Project Description

A key concerns in Canada and around the globe is that the generative AI (GenAI) is posing a real and significant existential risk to democratic discourse and, as a result, to democratic societies. This project studies how and why Canadians are vulnerable to the GenAI interference in digital democratic discourses and investigates advances in technology to make Canadians less vulnerable. ​

The project focuses on studying the ability of AI-driven chatbots to masquerade as humans on digital interactive public forums, and investigating technological countermeasures to mitigate this risk. To understand this vulnerability thoroughly, the project explores the interplay between the long-term user characteristics and other factors, such as user transient traits/circumstances, characteristics of the chatbots, the topic sensitivity, as well as the degree of topics’ importance/criticality for the participants. ​

Based on these findings, new ways to mitigate the risks are proposed and evaluated.

Project Description

Smart technologies are essential in hospitals and factories so securing the networks that connect them is critical. Devices such as wireless patient monitors, infusion pumps, and industrial sensors communicate using protocols like Zigbee, Z-Wave, Bluetooth, and Wi-Fi; each with its unique vulnerabilities that can expose systems to cyberattacks. ​

This project develops an artificial intelligence (AI)-powered cybersecurity system designed to detect and prevent attacks across diverse smart environments. Unlike traditional systems that rely onpre-defined rules, this solution will use advanced machine learning to recognize suspicious behavior and new threats as they emerge. We will build two realistic testbeds, one for Healthcare IoT (HIoT) and one for Industrial IoT (IIoT), to simulate real-world network conditions, capture traffic data, and test the system’s ability to identify intrusions.​

The final outcome will be an adaptable, explainable AI security model capable of operating across technologies and sectors.

Project Description

As electric vehicles (EVs) become more advanced and connected, they’re starting to do more than just drive us from place to place they are also becoming part of our energy systems. EVs can now send power back to the grid or even power homes, which helps save energy and reduce pressure on electricity networks. However, with this new connectivity comes new risks. These vehicles rely on smart systems to manage their batteries, and those systems are increasingly exposed to cyber threats like hacking or data breaches, which could affect not only the vehicle but also the energy grid. ​

This project builds a smarter, safer battery system called a Cyber-Resilient Battery Management System (CR-BMS). This system will combine artificial intelligence, secure communication, and blockchain technology to protect EVs from cyberattacks while safely managing energy flow between the car, home, and grid. The result will be a more secure and reliable energy system, helping to protect people, vehicles, and infrastructure. 

Project Description

This project addresses a critical gap in the cybersecurity of semi-autonomous vehicles—specifically, the exploitation of human behavioral patterns by adversarial actors. While Level 2Advanced Driver Assistance Systems (ADAS) are becoming standard in modern vehicles, they depend on the human driver as the ultimate fallback. This dependency creates a novel class of vulnerabilities that current cybersecurity frameworks fail to adequately address—those rooted in driver behavior, attention, and trust in automation.​

This project will develop a game-theoretic model capable of simulating adversarial strategies that exploit specific driver traits, such as delayed reaction times or overreliance on automation to close this gap. It will experimentally evaluate the effectiveness of Takeover Request (ToR) protocols in scenarios involving security breaches and silent system failures, accounting for cognitive workload and situational complexity. The resulting insights will inform the design of a risk-sensitive, driver-adaptive ToR mechanism.​

Finally, the project will construct a comprehensive threat assessment and remediation framework tailored to human-centric attack vectors, aligned with ISO/SAE 21434:2021 standards.​

Project Description

Modern cryptography is the foundation for practice and study of techniques to secure communication in the presence of adversarial behaviour. Many of these enable applications of great interest to our society including cloud computing, blockchains, anonymous verifiable voting, privacy-preserving machine learning, and more. ​

One key component to several proposed cryptosystems for these applications is the existence of groups of unknown order, that is, mathematical structures with a finite number of elements such that the exact number cannot be computed efficiently. Of particular interest are those that can be initialized with a trustless setup, where no third party knowing secret information that could be used to subvert the protocol is required. These concepts are relatively new in cryptography so there are several open problems related to their suitability for practical applications in terms of both security and efficiency. ​

The primary goal of this project is to obtain convincing evidence of the security and performance advantages of the underlying mathematical structures and algorithms enabling applications of trustless groups of unknown order, as well as a related mathematical structure called the infrastructure that may offer improved security and efficiency.​

Project Description

Malicious software impersonates users, penetrates business organizations through their devices, and locks user data to demand ransom. Improving the quality and security of software is a key theme in the national cybersecurity strategies of several countries, including Canada. There are also numerous efforts to detect malicious software led by industry. ​

Existing program-analysis- and machine-learning-based techniques have challenges to identify malicious mobile software. These include (a) the difficulty to identify representative features that accurately characterize software behaviors, (b) the highly imbalanced nature of training data which cannot be solved with simple over-and under-sampling, and (c) the limited reliability and explainability of machine-learning-based approaches, which often identify statistical correlations of low semantic value and thus fail to generalize and are prone to attacks. ​

This project investigates data, feature, and classifier properties that enable the development of reliable, explainable, and semantic malware detection techniques.

Project Description

This project investigates the privacy-invasivedata collection practices of commonly used SDKs. We develop methodologies and tools to inform users, developers, and enforcement agencies about preventing malicious activity and harmful data collection practices within the mobile ecosystem. ​

Studies are conducted of a large number of popular apps that request location permissions. The study focuses on measuring both observable and opaque network traffic to identify location-based data collection practices.​

Forensic analyze of SDKs determine whether they can access location data and under what conditions this access occurs.​

Comparison of the stated practices in app privacy policies with the actual data flows observed during app usage, allows for an assessment of their alignment with user privacy expectations.​

Project Description

​Canada’s critical infrastructure—including energy, water, transportation, and communications systems—is increasingly vulnerable to sophisticated cyber threats powered by artificial intelligence (AI) and, in the near future, quantum computing. Traditional security tools can no longer keep pace with the complexity, scale, and speed of these emerging threats. This project aims to fundamentally transform how we protect our most essential systems by developing the world’s first Zero-Touch cybersecurity framework powered by quantum-enhanced AI.

Led by a multidisciplinary team of experts in cybersecurity, AI, quantum computing, and industrial systems, this project will design an intelligent defense platform that can autonomously detect, analyze, and neutralize cyber threats in real time—before they cause harm. It will simulate future quantum-powered attack scenarios using advanced quantum generative models, enabling infrastructure to be “future-proofed” against evolving adversaries. Key innovations include a quantum-AI threat mapping engine, a root cause analysis system for self-healing, and tools for zero-day detection using quantum machine learning.

In addition to cutting-edge technology development, the project will train a new generation of quantum-AI security experts through hands-on experience with real-world ICS/OT systems. Industry partners—including Quantum City, IoTech Lab, and Axiam Technology—will provide vital infrastructure, testing environments, and commercialization pathways. Outputs will include open-access tools, policy guidance, and patentable technologies that advance Canada’s leadership in quantum-era cybersecurity.

By uniting the predictive power of AI with the computational edge of quantum algorithms, this initiative marks a major leap toward resilient, autonomous, and ethical cybersecurity systems. The project is not only defending today’s infrastructure but laying the foundation for a secure digital future.

Project Description

Quantum computing poses significant cybersecurity risks to critical infrastructures like power systems. Many conventional cryptographic methods could be broken by quantum computers, threatening the safety and stability of power grids. Despite efforts to develop Post-Quantum Cryptographic (PQC)standards, power systems have been slow to adoptquantum-safe practices due to the unique challenges of operational technology (OT). OT systems prioritize real-time performance, low latency, and high availability, while OT devices, with lifecycles of 10 to 20 years, are limited in computational power, making it difficult to implement advanced cryptographic solutions.​

This project evaluates how to integrate PQC algorithms into power systems without compromising performance. Using hardware-in-the-loop simulations, we ensure the solutions are practical for utilities and manufacturers. The project identifies PQC algorithms suitable for OT systems and provides a foundation for standardizing their use across critical infrastructures. It also suggest a transition plan to guide utilities in migrating safely to quantum-safe solutions.​