AI in Border Enforcement vs Border Protection: Two Diverging Uses of the Same Technology
The same sensor, two opposite purposes
A thermal camera at a maritime border can be used to intercept and push back a boat in violation of non-refoulement obligations, or to identify a vessel in distress and trigger a rescue. A facial-recognition system at a land crossing can be used to deny entry to an asylum seeker or to expedite family reunification. The technology is indifferent; the institutional purpose is everything. In 2026 the same AI stack is being deployed for both ends by different actors and, sometimes, by the same actor in different moments.
The enforcement stack in 2026
Border-enforcement deployments have matured around three capability classes. Surveillance: integrated radar, thermal, optical, and acoustic sensor networks feeding ML-based detection and tracking, deployed at sea by Frontex and national coast guards, and on land in the US-Mexico, Greek-Turkish, and Polish-Belarusian border zones. Biometric identification: large-scale facial and fingerprint matching against watchlists and prior-encounter databases. Risk scoring: algorithmic triage of travellers and asylum seekers, with iBorderCtrl as the most-discussed European pilot. Predictive routing: ML forecasts of migration flows used to position enforcement resources.
The protection stack uses many of the same components
Protection-side deployments include: maritime distress detection and rescue coordination (the UNHCR-IOM-IFRC missing migrants work draws on similar sensor inputs); biometric registration for refugees as documented above; chatbots and information services as documented above; and family-reunification matching using facial recognition with explicit consent. The technical components overlap heavily with enforcement; the differences are in governance, oversight, and purpose.
The legal framework is converging but uneven
The EU AI Act, in force from 2024 with staged application, classifies several border-enforcement uses as high-risk and prohibits some (real-time biometric identification in public spaces with narrow exceptions; social scoring; emotion recognition in certain contexts). The UN Special Rapporteur on the human rights of migrants has called for moratoria on several border-AI uses. The Council of Europe Framework Convention on AI extends related obligations across signatory states. Implementation lags the law in most jurisdictions.
The non-refoulement question
The principle of non-refoulement — that no one shall be returned to a place where they face persecution — is the central legal constraint on border-enforcement AI. Systems that identify, classify, and intercept asylum seekers before they can lodge a claim raise direct non-refoulement concerns even when each individual component is technically defensible. The UNHCR Note on the Externalisation of International Protection and the European Court of Human Rights jurisprudence on pushbacks define the legal envelope; AI deployments that have not been assessed against this envelope are operating in a contested space.
What to look for in any border-AI deployment
- What is the stated purpose, and what is the actual operational use?
- What is the legal basis under domestic and international law?
- What is the oversight mechanism, and is it independent of the deploying agency?
- What are the exclusion and false-positive rates, and how are they audited?
- Is there a documented complaint and remedy pathway accessible to those affected?
- Has a fundamental-rights impact assessment been conducted and published?
Further reading and primary sources
- EU AI Act: https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai
- Frontex: https://www.frontex.europa.eu/
- UN SR on migrants: https://www.ohchr.org/en/special-procedures/sr-migrants
- Council of Europe AI Convention: https://www.coe.int/en/web/artificial-intelligence
- IOM Missing Migrants Project: https://missingmigrants.iom.int/
- UNHCR Refworld: https://www.refworld.org/
Keep reading
Mediterranean Refugee Crossings in 2026: Routes and Risks
A 2026 explainer on Mediterranean refugee crossings, covering the Central, Eastern, and Western routes, drawing on UNHCR, IOM Missing Migrants, and Frontex reporting.
The Risks of AI in Humanitarian Work: Bias, Privacy, and Accountability (2026)
AI tools are now woven through humanitarian operations. The benefits are real and so are the risks. A frank look at the bias, privacy, and accountability gaps shaping the sector in 2026.
Who Is Responsible When AI Gets It Wrong in a Refugee Crisis? (2026)
When a model misclassifies a protection case or a biometric system locks a refugee out of food assistance, accountability is rarely clear. A close look at how responsibility is distributed in 2026 and where the gaps sit.
Can AI Be Neutral? The Problem of Bias in Humanitarian Data (2026)
AI systems learn from data that reflects who was easy to count and easy to reach. In humanitarian work, those gaps map directly onto vulnerability. A practical look at why AI cannot be neutral and what to do about it in 2026.
Data Privacy for Displaced People: What AI Systems Are Collecting in 2026
Biometrics, location, family ties, medical records, and protection flags all flow through AI systems used by humanitarian agencies. A clear look at what is being collected, who can see it, and what the risks are for displaced people in 2026.
AI and the Do No Harm Principle: Can They Coexist? (2026)
Do No Harm is the foundational principle of humanitarian action. AI systems introduce new categories of risk that the original framing did not anticipate. A practical assessment of whether the two can be reconciled in 2026.