Artificial intelligence (AI) has lately proved to be a coin with two sides. On the one hand, it can be leveraged as a powerful defensive mechanism to improve system preparedness and response against cyber incidents and attacks, and on the other hand, it can be a formidable weapon attackers can use to damage, compromise or manipulate systems. AI4CYBER ambitions to provide an Ecosystem Framework of next-generation trustworthy cybersecurity services that leverage AI and Big Data technologies to support system developers and operators in effectively managing robustness, resilience, and dynamic response against advanced and AI-powered cyberattacks. The project will deliver a new breed of AI-driven software robustness and security testing services that significantly facilitates the testing experts work, through smarter flaw identification and code fixing automation. Moreover, the project will provide cybersecurity services for comprehension, detection and analysis of AI-powered attacks to prepare the critical systems to be resilient against them. Incident response support by AI4CYBER will offload security operators from complex and tedious tasks offering them mechanisms to optimize the orchestration of the most appropriate combination of security protections, and continuously learn from system status and defences’ efficiency. The AI4CYBER framework will ensure fundamental rights and values-based AI technology in its services, through the integration of demonstrable explainability, fairness and technology robustness (security) capabilities in the AI4CYBER components. The ecosystem will be validated in three scenarios: i) Detection and Mitigation of AI-powered Attacks against the Energy Sector, ii) Robustness and autonomous adaptation of Banking applications to face AI-powered attacks and iii) Resilient hospital services against advanced and AI-powered cyber-physical attacks.

Collecting and analysing large amounts of data in the Cloud-to-Edge computing continuum raises novel challenges. Processing all this data centrally in cloud data centres is not feasible anymore as transferring large amounts of data to the cloud is time-consuming, expensive, degrade performance and may raise security concerns. Therefore, novel distributed computing paradigms, such as edge and fog computing emerged to support processing data closer to its origin. However, such hyper-distributed systems require fundamentally new methods. To overcome the limitation of current centralised application management approaches, Swarmhestrate will develop a completely novel decentralised application-level orchestrator, based on the notion of self-organised interdependent Swarms. Application microservices are managed in a dynamic Orchestration Space by decentralised Orchestration Agents, governed by distributed intelligence that provides matchmaking between application requirements and resources, and supports the dynamic self-organisation of Swarms. Knowledge and trust, essential for the operation of the Orchestration Space, will be managed through blockchain-based trusted solutions using methods of Self-Sovereign Identities (SSI) and Distributed Identifiers (DID). End-to-end security of the overall system will be assured by utilising state-of-the-art cryptographic algorithms and privacy preserving data analytics. Due to the imminent complexity of the decentralised system, novel simulation approaches will be developed to test and optimise system behaviour (e.g., energy efficiency) in the early stages of development. Additionally, the simulator will be further extended into a digital twin running in parallel to the physical system and improving its behaviour with predictive feedback. The Swarmchestrate concept will be prototyped on four real-life demonstrators from the areas of flood prevention, parking space management, video analytics and a digital twin of natural habitat.

With AI-enhanced components being deployed everywhere, including the very toolchains used for secure software development, the traditional security focus on software and hardware assets can no longer guarantee “secure services, processes and products, [and] digital infrastructures” in the EU Strategic Plan 2021-2024. Sec4AI4Sec wants to develop security-by-design testing and assurance techniques for AI-augmented systems, their software and AI assets. AI-augment systems provide an opportunity to democratize security expertise and give access to intelligent, automated secure coding and testing, by enabling novel capabilities, lowering development costs and increasing software quality (AI4Sec). They are also a risk: AI-augmented systems are vulnerable to new security technical threats specific to AI-based software, in particular where matters of fairness or explainability are important (Sec4AI). Sec4AI4Sec addresses these challenges to the fullest extent: “AI for better security, security for better AI.” The Sec4AI4Sec project will address these two facets of AI to achieve a deep scientific, economic and technological impact, while contributing to addressing key societal issues. It will validate its approach on three key scenarios of the EU Digital Compass towards Digital Sovereignty: 5G core virtualization, Autonomy for safety systems in aviation and security and Quality for 3rd party software assessment and certification. Sec4AI4Sec assembled a team with 5 leading Universities (Amsterdam, Cagliari, Hamburg, Lugano, Trento), 2 innovative SMEs (FrontEndART, Pluribus One), 3 Large Enterprises (Airbus, SAP, Thales) and 1 Center for digital innovation (Cefriel). The project will generate a set of innovative techniques and open-source tools, new methodologies for secure design and certification of AI-augmented systems, as well as reference benchmarks that can be used to standardize the assessment of research results in the secure software research community.

Recent reports state that the adoption of open-source software (OSS) helps, resulting in savings of about $60 billion per year to consumers. However, the use of OSS also comes at enormous cost: choosing among OSS projects and maintaining dependence on continuously changing software requires a large investment. Deciding if an OSS project meets the required standards for adoption is hard, and keeping up-to-date with an evolving project is even harder. It involves analysing code, documentation, online discussions, and issue trackers. There is too much information to process manually and it is common that uninformed decisions have to be made with detrimental effects. CROSSMINER remedies this by automatically extracting the required knowledge and injecting it into the IDE of the developers, at the time they need it to make their design decisions. This allows them to reduce their effort in knowledge acquisition and to increase the quality of their code. CROSSMINER uniquely combines advanced software project analyses with online monitoring in the IDE. The developer will be monitored to infer which information is timely, based on readily available knowledge stored earlier by a set of advanced offline deep analyses of related OSS projects. To achieve this timely and ambitious goal, CROSSMINER combines six end-user partners (in the domains of IoT, multi-sector IT services, API co-evolution, software analytics, software quality assurance, and OSS forges), along with R&D partners that have a long track-record in conducting cutting-edge research on large-scale software analytics, natural language processing, reverse engineering of software components, model-driven engineering, and delivering results in the form of widely-used, sustainable and industrial-strength OSS. The development of the CROSSMINER platform is guided by an advisory board of world-class experts and the dissemination of the project will be led by The Open Group.

Cancer patients (2.7M in Europe) with a positive prognosis are exposed to a high incidence of secondary tumours (≈1M). Bone metastases spread to the spine in 30-70% cases, reducing the load bearing capacity of the vertebrae and triggering fracture in 30% cases. Clinicians have only two options: either operate to stabilise the spine, or leave the patient exposed to a high fracture risk. The decision is highly subjective and can either lead to unnecessary surgery, or a fracture significantly affecting the quality of life and cancer treatment. The standard-of-care to classify patients with vertebral metastasis are scoring systems based on radiographic images, with little consideration of the local biomechanics. Current scoring systems are unable to establish an indication for surgery in around 60% of cases. Thus, there is an unmet need to accurately and timely quantify the risk of fracture to improve patient stratification and identify the best personalised treatment. This interdisciplinary project will develop Artificial Intelligence (AI)- and Physiology-based (VPH) biomechanical computational models to stratify patients with spine metastasis who are at high risk of fracture and to identify the best personalised surgical treatment. After rigorous model training with clinical (2000 retrospective cases) and biomechanical (120 ex vivo specimens) data, the new approach will be tested in a multicentric prospective observational study (200 patients). The models will be combined in a decision support system (DSS) enabling clinicians to successfully stratify metastatic patients. The models and the DSS will be designed so as to be suitable for regulatory requirements and future exploitation. METASTRA will propose new guidelines for the stratification and management of metastatic patients. METASTRA approach is expected to cut the uncertain diagnoses from the current 60% down to 20% of cases. This will reduce patient suffering, and allow cutting expenditure by 2.4B€/year.