The CYBER-BRIDGE project: An AI-Powered Platform for Cross-Border Cybersecurity, Compliance and Forensics in the EU

Nikolaos Kekatos, Georgios Koutidis, Charalampos Bratsas, Stefan Bucur, Alexandra-Mihaela Chirlejan, Mihaela Curcă, Angelos Fountoulakis, Michael Ioannou, Paul Lacatus, Alexios Lekidis, Emilios Charalambous, Avgi Michael, Tom Nianios, Simona Purcaru, Antonis VoulgaridisarXiv 2026 · arXiv

TL;DR

The CYBER-BRIDGE project introduces a unified, AI-powered platform designed to close the "structural resilience gap" in European cybersecurity. By binding cybersecurity operations, regulatory compliance, and digital forensics to a single hash-chained evidence layer, the platform ensures that a single incident record can simultaneously serve as a threat indicator, a compliance audit item, and a legally admissible forensic pack.

What Problem Does This Paper Address?

Cybersecurity in the EU is increasingly cross-border, yet the tools used to manage it remain siloed. An organization facing an attack must navigate a complex web of overlapping regulations: NIS2 for incident reporting, the Cyber Resilience Act (CRA) for product security, GDPR for data privacy, and the Law Enforcement Directive (LED) for forensic evidence.

Currently, security analysts often have to "re-key" the same data into different tools for threat sharing, compliance reporting, and forensic preservation. This fragmentation leads to missed regulatory deadlines, inconsistent data, and evidence that may not hold up in cross-jurisdictional legal proceedings. CYBER-BRIDGE tackles this by integrating these three domains—Cybersecurity, Compliance, and Forensics—into a single architectural framework.

Methods

The authors propose a modular, Docker-orchestrated microservice architecture (see Figure 2 below) built on five capability tiers: Presentation, Application, Intelligence, Security Assessment, and Data.

The CYBER-BRIDGE reference architecture showing the five-tier stack from UI to the PostgreSQL/pgvector data layer.

Key technical components include:

  • Shared Hash-Chained Evidence Layer: Every finding is wrapped in a custody envelope using SHA-256 hashes and RFC 3161 trusted timestamps. This ensures data integrity from the moment of intake.
  • AI-Driven Advisory: The platform utilizes Retrieval-Augmented Generation (RAG) powered by a local, on-premises Phi-4 model. This assistant helps analysts map incidents to regulatory requirements and drafts remediation roadmaps without leaking sensitive data to the cloud.
  • Automated Orchestration: It integrates standard security tools (Nmap, OWASP ZAP, Semgrep, etc.) under a single orchestrator, mapping all findings to the MITRE ATT&CK framework.
  • Cyber-Range: A sanitized replica of the backend image is used for upskilling personnel, allowing them to practice incident response in a controlled environment.

Datasets

As this is an architectural and systems paper, it does not use a single training dataset. Instead, its "data" consists of:

  • Regulatory Corpora: Seeded framework data from NIS2, CRA, GDPR, and ISO/IEC 27001 used to ground the RAG-based AI assistant.
  • Sectoral Pilots: The platform is being validated through five diverse use cases: Critical Energy Infrastructure, SME Security Service Providers, National CSIRTs, Law Enforcement Agencies, and IoT Manufacturers.

Key Insights and Ideas

The paper’s most compelling "aha moment" is the concept of "one artefact, three semantics." The authors argue that the difficulty isn't building the tools, but reconciling the conflicting requirements of different regulations on the same piece of data:

  • Tension between GDPR and Forensics: GDPR demands data minimization and redaction, while forensics requires complete, immutable records. CYBER-BRIDGE resolves this by separating the integrity anchor (the hash and timestamp) from the content. This allows a record to remain verifiable even after PII has been lawfully minimized.
  • Unified Regulatory Clocks: A single incident can trigger multiple countdowns (e.g., the 24h/72h NIS2 deadlines vs. the 72h GDPR deadline). The platform anchors these to a single evidence record to automate notification workflows.
  • Cross-Border Trust: By attaching explicit export metadata (TLP markings, legal basis), evidence packs can be verified by partners (like a foreign CSIRT or law enforcement) offline, without them needing direct access to the CYBER-BRIDGE platform.
Fig. 1. CYBER-BRIDGE platform overview: end-user roles (tier A) operate the three platform pillars plus the Cyber-Range (tier B), which produce the audit-grade outcomes (tier C), all bound to the shared hash-chained evidence layer.

Key Results

The paper defines quantitative targets for the project's success (KPIs), as summarized in the table below.

Platform-level pilot targets, including requirements for scanner integration, AI latency, and forensic evidence items.

While full performance metrics are slated for future deliverables, the authors position CYBER-BRIDGE as a unique "all-in-one" solution compared to existing open-source alternatives like MISP or TheHive, which typically focus on only one of the three pillars (see Table I in the paper).

Limitations and Future Work

The authors acknowledge several hurdles:

  • AI Trustworthiness: While RAG reduces hallucinations, they remain a possibility. Human-in-the-loop review is still essential for regulatory filings.
  • Interoperability: Different EU Member States may implement NIS2 differently, requiring the platform's "authority directory" to be highly configurable.
  • Trust Services: The current pilot uses FreeTSA for timestamping, which is not eIDAS-qualified. Future work involves integrating qualified trust service providers to meet the highest legal standards for machine-verifiable certificates.
  • Next Steps: Future papers will report on the actual KPIs from the five sectoral pilots and the accuracy/latency of the AI components.

Summary updated

Abstract

Cross-border cyber attacks on European SMEs and critical-infrastructure operators expose a structural cyber-resilience gap: cybersecurity (threat sharing, detection, response), compliance (framework assessment, certification) and digital forensics (cross-jurisdictional evidence) are jointly regulated by NIS2, the CRA, GDPR and the LED, yet their tooling evolved in silos and rarely interoperates across Member States. The EU-funded CYBER-BRIDGE project integrates the three domains under one architecture and a hash-chained evidence layer: AI-driven CTI sharing with ENISA-aligned notification, multi-framework assessments with retrieval-augmented advisory, and RFC 3161-anchored custody with lawful GDPR/LED exchange, all aligned with MITRE ATT&CK. A sanitised replica of the back-end image hosts a Cyber-Range for upskilling. The contribution is not the individual tools but the reconciliation mechanism that lets one incident produce, from a single evidence record, a cross-border CTI advisory, the regulatory notifications its deadlines demand and a re-verifiable forensic pack; we set out why binding three differently regulated evidence regimes onto one record is non-trivial and how the shared evidence layer resolves the resulting tensions. This paper introduces the project's scope, requirements, pilots, architecture and toolkits; quantitative KPIs follow in later deliverables.
BibTeX
@misc{kekatos2026cyber,
  author = {Nikolaos Kekatos and Georgios Koutidis and Charalampos Bratsas and Stefan Bucur and Alexandra-Mihaela Chirlejan and Mihaela Curcă and Angelos Fountoulakis and Michael Ioannou and Paul Lacatus and Alexios Lekidis and Emilios Charalambous and Avgi Michael and Tom Nianios and Simona Purcaru and Antonis Voulgaridis},
  title = {{The CYBER-BRIDGE project: An AI-Powered Platform for Cross-Border Cybersecurity, Compliance and Forensics in the EU}},
  year = {2026},
  eprint = {2611.21497},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2611.21497},
}
Figure