18 – 21 November 2025 | SEMICON Europa 2025| 
Munich, Germany
18 – 21 November 2025 | SEMICON Europa 2025| Munich, GermanyLorem ipsum dolor sit amet, consectetur adipiscing elit. Ut elit tellus, luctus nec ullamcorper mattis, pulvinar dapibus leo.
17-18 November 2025 | 3nd Plenary Meeting| Nicosia, Cyprus
17-18 November 2025 | 3nd Plenary Meeting| Nicosia, CyprusThe GuardAI consortium successfully wrapped up its 3rd Plenary Meeting in Cyprus after two productive days of discussions, reviews, and planning.
🔹 Key outcomes from the meeting:
• Strong technical progress across all work packages, with updated results on adversarial-robust perception and multi-agent autonomy.
• First steps toward integrating core GuardAI components into a unified system pipeline.
• Clear alignment on upcoming deliverables, demonstrations, and the roadmap leading to the mid-term review.
• Further exploration of synergies with other EU projects, strengthening cross-project collaboration and knowledge exchange within the cybersecurity and AI ecosystems.
A big thank you to all partners for the constructive exchanges and steady commitment to advancing secure and resilient AI for safety-critical applications.
GuardAI at European Researcher’s Night 2025
Our team showcased how AI systems can be deceived with adversarial attacks — including a live demo of person detection with an adversarial patch — highlighting why resilience and security in edge AI are critical.
We also presented demo videos illustrating GuardAI’s vision for secure and resilient AI at the edge.
It was inspiring to see so much curiosity and discussion around robust AI — exactly the awareness GuardAI aims to raise. More to come!
1st Promotional Video
🚀 We are excited to share the first GuardAI project video!
GuardAI is building the foundations for trustworthy and certifiable AI, developing innovative solutions to protect safety-critical applications such as autonomous vehicles, drones, and next-generation cybersecurity.
📽️ Watch the video to learn more about our vision and goals: Link
1st Newsletter
🚀 We are excited to share the very first newsletter of the project, highlighting our journey, progress, and insights towards building resilient, trustworthy, and secure AI systems for the future.
🚨 New Publication Alert from the GuardAI Project! 🚨
We’re thrilled to share that our paper “FAST AND ACCURATE OUTLIER-AWARE LIDAR SUPER-RESOLUTION FOR SLAM APPLICATIONS” has been accepted to ICIP 2025 — the IEEE International Conference on Image Processing, taking place in Anchorage, Alaska! 🌲❄️
🔍 What’s it about?
We tackle the challenge of enhancing low-resolution LiDAR for SLAM by introducing a Deep Unrolling-based Super-Resolution model that’s both accurate and fast. Our method integrates a novel outlier removal module to ensure structural integrity and real-time performance, all while minimizing computational overhead.
📈 Within a full SLAM pipeline, our approach significantly improves pose estimation accuracy and efficiency, outperforming existing SR methods.
🚨 New Publication Alert from the GuardAI Project! 🚨
We are proud to share that the paper titled “Advancing B5G Security: An AI-Augmented Intrusion Detection System using a Real-Time Attack Generator” has been accepted for presentation at the 2025 IEEE CSR Workshop on Cyber-Physical Resilience and Security Against Digital Breakdowns (CYPRES) this August!
🔍 What’s inside the paper?
✅ A novel real-time Attack Generation Engine (AGE) simulating critical threats like gNodeB DoS attacks, NAS message tampering, and UPF exploitation
✅ Deployment and evaluation on the CERTH 5G testbed, providing realistic threat emulation
✅ Integration with advanced AI-based Intrusion Detection Systems (IDS) using LSTMs, GNNs, and Transformers for high-fidelity detection
🔗 This work is also a result of synergies across EU-funded initiatives, building on the Natwork Project and feeding directly into the technical foundations of GuardAI.
31 July 2025 | AI Act Workshop
As part of our commitment to responsible and trustworthy EdgeAI, the GuardAI project hosted an internal workshop on the EU AIAct and its implications to the technologies developed within the project.
The session offered: A clear overview of the AI Act’s structure, terminology, and risk-based approach
A practical reflection on how the regulation applies to key use cases within GuardAI (drones, autonomousvehicles, and 5Gnetworks)
Insights on legal roles and responsibilities, risk classification and management, and compliance challenges
The workshop concluded with an open discussion, on how to better align with the provisions of the AI act and navigate the evolving regulatory landscape
A big thank you to our partners from UNIVIE for organizing and leading this insightful session! 
July 2025 | 17th Cyprus Workshop on Signal Processing (CWSPI)
Our partners from the KIOS Research and Innovation Center of Excellence at the University of Cyprus, represented by Research Lecturer Christos Kyrkou, presented ongoing work from the GuardAI project at the 17th Cyprus Workshop on SignalProcessing and Informatics in Nicosia. The presentation focused on the use of distilled and fine-tuned denoising diffusion models for adversarial purification, aiming to enhance the robustness of edge-based vision systems against adversarial attacks.
06 June 2025 | 2nd Standardisation Webinar
The Cyprus Organisation for Standardisation (CYS) organized the second session of our internal webinar series on Standardization in Artificial Intelligence, focusing on engagement with technical committees.
Key topics:
✅ European & International Technical Committees (CEN/CLC, ETSI, ISO)
✅ Harmonized Standards & the AI Act
✅ Pre-normative documents (CWAs)
✅ Practical ways of engagement
This series is part of our mission to enhance the robustness and trustworthiness of Edge AI systems. Stay tuned as we continue building the foundations for secure and resilient AI!
🗓️ 10-11 April 2025 | 2nd Plenary Meeting| 📍 Athens, Greece
The GuardAI consortium successfully concluded its 2nd Plenary Meeting in Athens — two days of focused collaboration, technical alignment, and strategic planning.
Key highlights from Day 1:
✅ Comprehensive updates on WP1 – Requirements, Use Case-Driven Methodology, and Architecture.
✅ Technical deep-dives into robustness, adversarial resiliency, and context-aware AI under WP2.
✅ Strategic planning for dissemination, exploitation, and certification (WP5) with active contributions from all the partners.
The day concluded with a well-deserved dinner in the heart of Athens 🍽️ — a great way to build synergies and celebrate progress!
Key highlights from Day 2:
🚀 Use Case Presentations – showcasing real-world Edge AI implementations and experiments
📊 Focused updates on Project Coordination, Risk Assurance, and Ethics (WP6)
🔍 Insightful planning for the upcoming 6 months: Deliverables, milestones, and critical risk mitigation strategies
🤝 Wrap-up discussions and alignment on next steps — we’re moving forward stronger and more connected
The engagement and collaboration during the meeting confirmed the consortium’s shared commitment to advancing the state of secure Edge AI for safety-critical domains.
Looking forward to the next milestones as we continue building a robust AI trust framework under Horizon Europe.
🗓️ 10-11 October 2024 | Kick-off Meeting| 📍 Nicosia, Cyprus
The GuardAI kick-off meeting took place on 10-11 of October at the University of Cyprus.
The 10 partners from across Europe met to initiate the ambitious goal of developing cutting-edge, secure, and robust, solutions tailored to the specific needs of edge AI safeguarding critical infrastructure and systems.
The partners will tackle challenging emerging issues of adversarial robustness and context awareness in modern AI systems and high-stakes domains such as drones connected and autonomous vehicles and 5G infrastructure.
