Objectives
Objective 1
Establish a comprehensive security and risk assessment process for the complete lifecycle of an AI model in a supply chain and beyond.
Objective 2
Provide a Security, Trust and Robustness (STR) Defence Framework offering holistic AI model security-by-design approach throughout its overall lifecycle and under any AI realization paradigm (single, collaborative learning or interconnected AI agents supply chain).
Objective 3
Introduce and support a new paradigm on Security, Trust and Robustness for AI models compliant to the EU approach on AI and propose the AI Model Bills of Material as a universal AI Model description structure.
Objective 4
Infuse into the overall AI model design operation continuum (design, training, deployment) context awareness mechanism that allow a deployed AI model to be aware of its constantly evolving cybersecurity environment and adapt to adversaries based on such a context.
Objective 5
Provide the foundation and create the momentum for an STR AI supply chain ecosystem that offers to the AI community the CoEvolution hub of secure, trusted and robust AI models, trusted training datasets as well as the framework to utilize them in various AI paradigms (collaborative learning) and the AI supply chain.
Objective 6
Integrate, validate and assess the CoEvolution overall solution in 2 complex pilots that feature all AI model realization paradigms (i.e. single, collaborative learning and Interconnected AI agents) showcasing the security of the overall AI supply chain.
Objective 7
Provide a viable and well-structured business plan on the exploitation of the CoEvolution solution that will lead into a sustainable exploitable outcome after the duration of the project.