The contemporary AI landscape demands a holistic framework ensuring security across the supply chain and entire AI lifecycle. Despite existing adversarial attack techniques, a comprehensive end-to-end flow for identifying threats and vulnerabilities with associated risks is lacking. The EU, through initiatives like the AI Act, emphasizes safety and trustworthiness in AI applications but lacks a system managing weaknesses in a networked AI-supply chain. The CoEvolution project integrates its architecture components to create an end-to-end Security, Trust, and Robustness (STR) assessment solution, generating context-aware AI models characterized by their AI Model Bill of Materials (AIMBOM). The goal is a universal hub providing a coherent STR risk assessment and security assurance flow, aligning with MLDevOps and EU AI regulatory frameworks. The paradigm includes novel AI model descriptions, AIMBOM management, security monitoring, and context awareness. CoEvolution introduces a new STR paradigm based on Bills-of-Materials, offering a unified approach to describing AI models in supply chains, ensuring STR compliance with EU directives on trust, fairness, data governance, and GDPR guidelines. Open source trusted datasets and CoEvolution-developed AI models enhance the hub's capabilities, aiming for a robust, adaptable risk analysis and security assessment framework aligned with evolving AI cybersecurity threats.

cPAID envisions researching, designing, and developing a cloud-based platform-agnostic defense framework for the holistic protection of AI applications and the overall AI operations of organizations against malicious actions and adversarial attacks. cPAID aims at tackling both poisoning and evasion adversarial attacks by combining AI-based defense methods (e.g., life-long semi-supervised reinforcement learning, transfer learning, feature reduction, adversarial training), security- and privacy-by-design, privacy-preserving, explainable AI (XAI), Generative AI, context-awareness as well as risk and vulnerability assessment and threat intelligence of AI systems. cPAID will identify guidelines to a) guarantee security- and privacy-by-design in the design and development of AI applications, b) thoroughly assess the robustness and resiliency of ML and DL algorithms against adversarial attacks, c) ensure that EU principles for AI ethics have been considered, and d) validate the performance of AI systems in real-life use case scenarios. The identified guidelines aspire to promote research toward developing certification schemes that will certify the robustness, security, privacy, and ethical excellence of AI applications and systems.

SYNAPSE aims to design, develop & deliver an Integrated Cyber Security Risk & Resilience Management Platform, with holistic Situational Awareness, Incident Response & Preparedness capabilities. The proposed platform will encompass: (i) Incident Response through process automation and orchestration mechanisms, also covering organisational/business aspects (e.g., business continuity processes); (ii) AI-enhanced Situational Awareness, encompassing extraction & analytics of actionable and pertinent Cyber Threat Intelligence (CTI), along with attack early warning & threat hunting systems; (iii) Preparedness through cybersecurity, privacy & business continuity training, covering different training delivery means, allowing it to tailor the delivery method to the content; (iv) Technical & economic risk management, integrating outputs of (i)-(iii) above and supporting risk-benefit analyses (including what-if scenarios) to inform decision-making and enable risk transfer schemes with Smart Contract-enabled cybersecurity insurance; (v) Continuous feedback between (i)-(iv) above, along with standards-based sharing, alerting & reporting (intra- & inter- Member State), based on outputs of (i)-(iii) above, thus enabling the establishment of shared situational awareness, coordinated response and joint preparedness

As Internet of Things (IoT) and IoT-Edge-Cloud continuum technologies advance, physical environments are becoming increasingly equipped with sensors, fuelling the development of smart space ecosystems. Massive quantities of data produced by IoT devices revolutionize the way such ecosystems operate via the exploitation of AI models/services. This has led to the emergence of the so-called Artificial Intelligence of Things (AIoT) systems. In general, designing techniques to promote robustness, efficiency and continual operation of AIoT systems requires realistic and trustworthy data at scale. However, such data is not always easy to obtain due to the cost of smart space construction, the inconvenience of long-term device tracking, the sensor/knowledge data gaps in diverse scenarios of a smart space, and the restrictions imposed on sensitive data sharing. Furthermore, an efficient AIoT system operation requires trustworthy AI services, as well as novel approaches for speeding up their inference across the IoT-Edge/Cloud continuum. PANDORA aims to devise and implement a comprehensive framework enabling the delivery of trustworthy datasets of smart space ecosystems, as well as the deployment and green operation of AIoT systems in such spaces. PANDORA spans two phases: (1) prior to AIoT system deployment; (2) post AIoT system deployment and operation. Phase 1 proposes and combines a series of novel techniques such as synthetic data generation, quantification of uncertainties, and data summarization for the delivery of trustworthy datasets, as well as explainable AI and domain-informed model training/testing in smart space ecosystems. Phase 2 defines novel AIaaS and CaaS techniques for the robust, explainable, green and continual operation of AIoT systems deployed in such spaces. The trustworthiness and applicability of the PANDORA framework will be tested through five pilot cases hosting AIoT applications in smart buildings, factories and critical infrastructures.

Artificial Intelligence (AI) is widely studied and finding increasing adoption across communication technologies spanning network layers and business ecosystems. It is anticipated to play a central role in the design and operation of future 6G networks. Despite the promise of AI, there remain many obstacles to its use in communication networks. The introduction of software defined elements such as radio access network (RAN) intelligent controllers (RIC) enables multi-party applications for the control and management of networks. However, AI functions are still nascent and such structures do not extend to optical networks or multi-controller environments. 6G-XCEL seeks to address these challenges through research on high edge network use cases that employ multi-party AI controls running over compute accelerators to coordinate control across radio and optical networks. It will develop a reference framework for AI in 6G that will pave the way towards global validation, adoption and standardisation of AI approaches. This framework will enable decentralised AI-based network controls across network domains and physical layers, while promoting security and sustainable implementations. Using the latest AI algorithms and data compression, research on the resulting decentralised multi-party, multi-network AI (DMMAI) framework will enable the development of reference use cases, data and model repositories, curated training and evaluation data, as well as technologies for its use as a benchmarking platform for future AI/ML solutions for 6G networks. 6G-XCEL will bring together a large ecosystem of researchers from the EU and US to implement elements of the DMMAI framework in their testbeds and labs, integrating it into their research programs and validating the framework across platforms. Working with standardisation groups within each jurisdiction, 6G-XCEL will achieve joint progress towards large scale application of AI in 6G networks.