USE of AI STANDARDS in BORDERLINK

Summary: BORDERLINK is positioning itself as one of the first Horizon Europe security projects to adopt a dual‑layer AI governance model: (1) the EU Living Guidelines on Responsible Use of Generative AI in Research (2026) and (2) its own Responsible & Explainable AI Standards, grounded in the EU AI Act, GDPR, OECD AI Principles, Council of Europe AI Convention, and ENACT law‑enforcement standards. This combination gives BORDERLINK a defensible, future‑proof AI governance framework.

How BORDERLINK Uses AI and and the Standards That Govern It

Artificial intelligence is reshaping border management, customs operations, and cross‑border security. But in sensitive environments like these, how AI is used matters just as much as what it does. The BORDERLINK project has taken a proactive stance: adopting a rigorous, multi‑layered framework to ensure that AI is used responsibly, transparently, and in full alignment with European legal and ethical expectations.

This blog explains how BORDERLINK uses AI, and which standards guide its approach, combining insights from the project’s internal protocol and the European Commission’s Living Guidelines on the Responsible Use of Generative AI in Research (May 2026).

A Dual Framework: Research Integrity + Operational AI Governance

BORDERLINK’s AI protocol is built on two complementary pillars:

-              EU Living Guidelines on Responsible Use of Generative AI; covering research integrity, transparency, privacy, reproducibility, and responsible use of generative AI.

-              BORDERLINK Responsible & Explainable AI Standards; covering operational AI use in border/customs contexts, risk mitigation, adversarial robustness, and law‑enforcement proportionality.

Together, they ensure that AI is used safely, ethically, and legally across both research and operational dimensions.

Legal & Ethical Alignment: AI Act + Fundamental Rights

BORDERLINK commits to full compliance with the EU AI Act, applying its risk‑based framework to all AI activities. This mirrors the EU guidelines’ emphasis on legality and rights protection: “Researchers respect applicable national, EU and international legislation… especially sensitive in relation to the protection of intellectual property rights and personal data.” (EU Guidelines, 2026)

The project also integrates: OECD AI Principles; Council of Europe AI Convention; and GDPR and proportionality principles.

This ensures that AI systems respect privacy, non‑discrimination, human dignity, and lawful processing, essential in border and customs environments.

Human‑Centric AI: Oversight, Agency, and Accountability

We insist that humans remain fully responsible for all AI‑assisted outputs. “Researchers are accountable for the integrity of the content they produce… AI systems are neither authors nor co-authors.” (EU Guidelines, 2026)

BORDERLINK operationalises this by:

-              Ensuring meaningful human oversight in all AI‑supported decisions;

-              Defining clear roles and responsibilities; and

-              Maintaining audit trails and compliance reporting.

This prevents over‑reliance on automated systems and ensures accountability in sensitive operational contexts.

Transparency & Explainability: No Black Boxes

Transparency is a shared cornerstone: “Researchers… detail which generative AI tools have been used substantially… disclose limitations and biases.” (EU Guidelines, 2026)

BORDERLINK extends this principle by requiring:

-              Documentation of data sources, model logic, and decision pathways;

-              Lifecycle explainability from design → training → deployment → evaluation; and

-              Open publication of non‑sensitive datasets and methodologies.

This builds trust with stakeholders and supports reproducibility.

Fairness, Bias Testing & Non‑Discrimination

Both frameworks emphasise fairness: “Researchers maintain a critical approach… aware of limitations such as bias, hallucinations and inaccuracies.” (EU Guidelines, 2026)

BORDERLINK applies this through:

-              Regular bias testing;

-              Monitoring for discriminatory outcomes in border/migration contexts; and

-              Ensuring proportionality and necessity in law‑enforcement use.

This is essential to prevent algorithmic discrimination in high‑stakes environments.

Privacy, Confidentiality & GDPR Compliance

The EU guidelines warn against uploading sensitive or unpublished data to external AI systems: “Researchers protect unpublished or sensitive work… by taking care not to upload it into an external AI system.” (EU Guidelines, 2026)

BORDERLINK reinforces this with:

-              Federated databases;

-              Privacy‑enhancing techniques;

-              Strict GDPR compliance; and

-              Controlled environments for AI processing.

This ensures that sensitive customs and border data remains protected.

Security, Resilience & Adversarial Robustness

The EU guidelines highlight cybersecurity and responsible data handling: “Researchers understand the technical, ethical and security implications… ensure appropriate cybersecurity of systems.” (EU Guidelines, 2026)

BORDERLINK aligns its activities with these standards by:

-              Employing systems to withstand adversarial attacks;

-              Preventing misuse and manipulation; and

-              Ensuring resilience in operational border environments.

This aligns with ENACT law‑enforcement standards, which emphasise proportionality, necessity, legality, and robustness.

Stakeholder Engagement & Open Science

We have a broad engagement with universities, research organisations and end-user stakeholders who cultivate responsible use of AI.

BORDERLINK operationalises this by involving customs authorities, law enforcement and NGOs, and by openly sharing non‑sensitive datasets and methodologies to support reproducibility and trust.

Environmental Responsibility

The EU guidelines highlight the environmental impact of AI: BORDERLINK incorporates this by selecting efficient AI models and monitoring resource use.

Conclusion: BORDERLINK as a Model for Responsible AI in Security Research

By combining the EU Living Guidelines with its own Responsible & Explainable AI Standards, BORDERLINK sets a new benchmark for responsible AI in border and customs innovation.

It demonstrates that high impact AI can be: transparent, explainable, fair, secure, legally compliant, and environmentally responsiblewhile still delivering cutting edge capabilities for cross border cooperation.