AI Border Control Systems and Digital Market Partitioning

Introduction

AI border control systems use artificial intelligence, biometrics, machine learning, facial recognition, risk-scoring, automated document verification, passenger-data analytics, and interoperable databases to control or facilitate movement across national or regional borders.

From a competition-law perspective, these systems create a distinctive risk: technology can transform geographical borders into digital market boundaries. A dominant platform, airport operator, identity provider, cloud provider, biometric vendor, or government-linked infrastructure operator may use technical architecture, data restrictions, interoperability barriers, licensing conditions, or algorithmic access rules to divide markets along territorial lines.

The central competition-law question is therefore:

When does legitimate territorial differentiation required for border security become unlawful digital market partitioning or exclusionary conduct?

The answer depends upon the legal jurisdiction, the market structure, the actor involved, the justification for the territorial restriction, and its effects on actual or potential competition.

I. Meaning of AI Border Control Systems

AI border control systems may include:

1. Automated Border Control (ABC) gates
2. Facial-recognition systems
3. Fingerprint and iris identification
4. AI-assisted passport verification
5. Passenger Name Record (PNR) analysis
6. Advance Passenger Information (API) systems
7. Automated risk scoring
8. Watch-list matching
9. Digital identity systems
10. Biometric databases
11. AI-assisted customs screening
12. Predictive cargo-risk systems
13. Cross-border immigration databases
14. Automated visa-processing systems
15. Airport security analytics.

These systems frequently depend upon several interconnected markets:

Identity data → biometric technology → AI software → cloud infrastructure → airport/border hardware → government databases → airlines/travel platforms → verification services

Consequently, competition problems may arise at multiple levels rather than in a single conventional product market.

II. Digital Market Partitioning

Digital market partitioning occurs when technological, contractual, regulatory, or commercial arrangements divide otherwise potentially integrated digital markets into separate territorial or customer segments.

Traditional market partitioning could involve:

• territorial sales restrictions;
• export bans;
• customer allocation;
• geographic pricing.

Digital partitioning can additionally occur through:

• IP/geolocation blocking;
• API restrictions;
• data localisation;
• incompatible technical standards;
• regional software versions;
• access-token restrictions;
• algorithmic routing;
• identity-system incompatibility;
• territorial licensing;
• cloud-region restrictions;
• discriminatory interoperability;
• regional exclusion from platforms.

AI makes the problem more difficult because partitioning may be embedded directly into the algorithmic architecture.

III. How AI Border Control Can Produce Market Partitioning

1. Algorithmic Geographic Blocking

An AI system may automatically determine that a user, airline, passenger, service provider, or transaction belongs to a particular territory.

For example:

Country A's identity-verification algorithm accepts only credentials issued within Country A.

If competing identity providers cannot obtain access to the verification infrastructure, the incumbent system can become a digital gateway controlling access to the market.

2. Data Localisation

A border-control authority may require biometric or passenger data to remain within a particular territory.

Some localisation requirements may be legitimate because of:

• national security;
• privacy;
• sovereignty;
• cybersecurity;
• law-enforcement requirements.

However, competition concerns may arise where localisation is designed or applied in a manner that unnecessarily excludes competing providers.

The competition-law inquiry therefore should distinguish:

genuine regulatory localisation

from

commercially motivated localisation that protects an incumbent.

IV. Interoperability Restrictions

AI border systems depend heavily upon interoperability.

A system may refuse to communicate with:

• foreign biometric databases;
• competing identity providers;
• alternative cloud platforms;
• rival AI verification systems;
• competing airport systems;
• third-party APIs.

If a dominant infrastructure provider controls the essential technical interface, refusal to provide interoperability may potentially constitute an exclusionary practice.

V. Territorial Licensing

A biometric or AI company may license its system only to selected operators in particular territories.

Territorial licensing is not automatically unlawful.

The competition issue becomes stronger where:

• the supplier is dominant;
• territorial restrictions prevent parallel trade;
• customers are allocated geographically;
• competing distributors are prevented from supplying neighbouring territories;
• the restriction fragments an integrated market.

VI. Algorithmic Discrimination Between Territories

AI systems may automatically impose different:

• prices;
• access conditions;
• verification requirements;
• transaction limits;
• API permissions;
• service quality;
• processing speeds

depending upon the user's jurisdiction.

This raises the possibility that a technological system is performing digital customer allocation without an explicit human agreement.

Competition authorities may therefore need to examine the underlying:

• code;
• training data;
• business rules;
• contractual instructions;
• optimisation objectives;
• enforcement mechanisms.

VII. Dominance and Border-Control Infrastructure

A competition problem is particularly significant where one undertaking controls a critical infrastructure layer.

For example:

Border AI platform → biometric database → API → airport systems → airlines → passengers

If the platform is indispensable for downstream competitors, the platform operator may acquire substantial gatekeeping power.

Relevant theories include:

A. Refusal to deal

A dominant operator refuses access to an indispensable interface.

B. Discriminatory access

The operator supplies its own downstream business more favourable access than competitors.

C. Self-preferencing

The border infrastructure gives preferential treatment to affiliated verification, identity, cloud, or analytics services.

D. Tying

Access to the border-control system is conditioned upon purchasing another service.

E. Exclusive dealing

Airports or governments are required to use only the dominant supplier.

F. Leveraging

Market power in border infrastructure is extended into adjacent digital markets.

VIII. Relevant Case Laws

Because AI border-control market partitioning is an emerging field, there is not yet a large body of reported judgments dealing specifically with AI border-control algorithms. The following cases provide important principles that can be applied by analogy.

1. Consten and Grundig v Commission

Case 56/64 and 58/64, Consten and Grundig v Commission (1966)

Principle

The European Court of Justice treated agreements designed to protect national territories from parallel trade as contrary to the fundamental competition principle against market partitioning.

Relevance to AI border systems

The case provides an important conceptual foundation for digital territorial partitioning.

A modern equivalent could involve:

• geo-blocking;
• territorial API restrictions;
• region-specific authentication;
• digital licensing walls.

The technological form may change, but the competition concern remains the creation of artificial territorial barriers.

2. Football Association Premier League Ltd v QC Leisure

Joined Cases C-403/08 and C-429/08

Principle

The Court examined territorial restrictions associated with broadcasting rights and found that contractual arrangements designed to eliminate cross-border trade could raise serious competition concerns.

AI relevance

The case is particularly relevant where digital technologies are used to create territorial exclusivity.

For example, an AI platform could technically prevent:

• a foreign supplier from providing verification services;
• cross-border authentication;
• international data processing;
• access to competing digital services.

The fact that the restriction is technically enforced rather than manually enforced does not eliminate its potential competition-law significance.

3. GlaxoSmithKline Services Unlimited v Commission

Joined Cases C-501/06 P, C-513/06 P, C-515/06 P and C-519/06 P

Principle

The litigation concerned pharmaceutical distribution arrangements and territorial restrictions.

The Court recognised that restrictions on parallel trade must be analysed carefully, including their competitive effects and the relevant legal context.

AI border-control relevance

This is useful for distinguishing:

• legitimate territory-specific regulatory requirements

from

• artificial restrictions designed to prevent cross-border competition.

An AI border-control provider cannot necessarily justify every territorial restriction merely by labelling it a "security requirement."

4. Pierre Fabre Dermo-Cosmétique

Case C-439/09, Pierre Fabre Dermo-Cosmétique (2011)

Principle

The Court considered a restriction effectively preventing internet sales and emphasised the importance of online channels for competition.

AI relevance

The case illustrates how a technological restriction can have competition consequences even where the restriction does not expressly say:

"Competitors are prohibited."

For AI border-control systems, comparable effects could arise where technical design prevents:

• remote verification;
• cross-border digital identity services;
• online access;
• foreign suppliers from reaching customers.

Thus, technical architecture can itself function as a competitive restriction.

5. Google Shopping

Google Search (Shopping), Commission Decision AT.39740 (2017), subsequently litigated before the EU Courts

Principle

The case concerned preferential treatment of Google's own comparison-shopping service within its general search infrastructure.

AI border-control relevance

The important analogy is self-preferencing through infrastructure control.

Suppose a dominant border-control platform operates:

1. the identity-verification infrastructure;
2. the API;
3. the AI matching system; and
4. an affiliated identity service.

If the system systematically gives the affiliated service preferential:

• ranking;
• verification speed;
• API access;
• authentication success;
• data availability,

competition concerns may arise.

The border-control system would effectively become both:

infrastructure operator + market participant.

6. Slovak Telekom v Commission

Joined Cases C-152/19 P and C-165/19 P

Principle

The case concerned access to infrastructure controlled by a dominant undertaking and the conditions under which refusal or restriction of access may constitute abusive conduct.

AI relevance

This provides a useful framework for AI border infrastructure.

Where an undertaking controls a critical:

• database;
• API;
• biometric verification network;
• identity platform;
• cloud infrastructure,

and downstream competitors cannot realistically operate without access, discriminatory or restrictive access conditions may become competition-law concerns.

7. Bronner v Mediaprint

Case C-7/97, Oscar Bronner GmbH & Co. KG v Mediaprint

Principle

The Court established a demanding test for when refusal to provide access to infrastructure can constitute abuse of dominance.

Among the relevant considerations is whether the facility is genuinely indispensable and whether duplication is practically or economically impossible.

AI border-control relevance

The case is highly useful for analysing whether a government-connected or private AI border platform constitutes an essential facility.

For example:

If a biometric verification API is the only realistic method through which airlines can obtain government-required authentication, access may become competitively significant.

But indispensability should not simply be assumed. A competition analysis must examine whether realistic alternatives exist.

8. Magill

Joined Cases C-241/91 P and C-242/91 P, RTE and ITP v Commission

Principle

The case developed the exceptional circumstances under which refusal to license intellectual property may constitute abuse of dominance.

AI relevance

Modern AI border-control systems may involve:

• proprietary algorithms;
• biometric databases;
• technical standards;
• APIs;
• copyrighted software;
• database rights.

If a dominant operator controls indispensable technology and refuses access, Magill provides part of the legal framework for examining whether the refusal crosses the competition-law threshold.

IX. Digital Market Partitioning Through Data

Data can become the most important partitioning mechanism.

Imagine:

Country A

biometric data → national database → AI verification

but

Country B

separate database → incompatible AI model → no interoperability.

If the databases are technically incompatible, competitors may need to develop separate infrastructure for every jurisdiction.

This can increase:

• entry costs;
• compliance costs;
• development costs;
• switching costs;
• network effects.

The result may be digital fragmentation.

X. Network Effects

AI border-control systems frequently exhibit strong network effects.

The more:

• airports;
• airlines;
• governments;
• passengers;
• identity providers

that participate, the more valuable the network becomes.

This can create a feedback loop:

More users → more data → better AI → greater accuracy → more adoption → more data.

Eventually, an incumbent system may become difficult for competitors to challenge.

This is particularly important when the incumbent also controls the technical standards.

XI. Switching Costs and Lock-In

Border systems are generally expensive to replace.

An airport or government may need to change:

• biometric hardware;
• software;
• databases;
• cybersecurity systems;
• APIs;
• staff training;
• certification;
• regulatory approvals.

Consequently, a dominant supplier can potentially impose significant switching costs.

Competition authorities may examine whether contractual or technical arrangements artificially increase those costs.

XII. Tying and Bundling

An AI border-control supplier might condition access to its biometric system upon purchasing:

• cloud hosting;
• cybersecurity;
• analytics;
• identity verification;
• passenger management software.

This could create a leveraging theory of harm.

The structure would be:

Market A: border-control AI

↓ leverage

Market B: cloud/identity/analytics services.

The relevant question would be whether the tying arrangement forecloses competitors in the second market.

XIII. Public Procurement Dimension

AI border systems are frequently purchased through public procurement.

Competition problems may arise when procurement specifications:

• unnecessarily require proprietary technology;
• favour an incumbent's architecture;
• exclude interoperable alternatives;
• make competing systems incompatible;
• bundle unrelated services;
• create long-term exclusive arrangements.

The competition analysis must be separated from legitimate public-security requirements.

A technically restrictive procurement condition is not automatically anticompetitive merely because it disadvantages some suppliers.

XIV. Government Action Versus Undertaking Conduct

An especially important distinction is whether the partitioning is created by:

1. Government regulation

or

2. Conduct of a private undertaking.

A state may impose territorial restrictions for:

• immigration control;
• national security;
• privacy;
• sanctions;
• customs;
• cybersecurity.

Such measures raise different legal questions from a private technology company deliberately creating territorial barriers to protect its commercial position.

Where government and private infrastructure operators cooperate, the analysis becomes more complex.

XV. Legitimate Justifications

Territorial differentiation can be legitimate where objectively justified.

Potential justifications include:

1. National security

Border-control information can involve highly sensitive security information.

2. Privacy

Biometric information may require territorial processing restrictions.

3. Cybersecurity

Governments may require data to be stored in specified environments.

4. Sovereignty

States may require sensitive identity systems to remain under domestic control.

5. Legal compliance

Different jurisdictions may impose different:

• immigration laws;
• biometric requirements;
• data-protection rules;
• sanctions;
• customs rules.

6. Technical security

Interoperability can itself create cybersecurity risks.

The competition question is therefore not simply:

"Does the system divide markets?"

but:

"Is the division necessary and proportionate to a legitimate objective, or does it unnecessarily exclude competition?"

XVI. Algorithmic Collusion and Border Markets

A further issue arises when several competing border-service providers use AI systems that monitor one another's:

• prices;
• capacity;
• contracts;
• access conditions.

Algorithms can facilitate rapid market coordination.

For example:

Supplier A's algorithm observes Supplier B's pricing and automatically adjusts its own price.

If algorithms independently reach parallel pricing outcomes, authorities must distinguish:

• lawful intelligent pricing;
• conscious coordination;
• communication between competitors;
• algorithmically facilitated collusion.

The existence of an AI system does not eliminate traditional cartel principles.

XVII. Digital Partitioning and International Trade

Artificial territorial restrictions can also intersect with:

• free movement principles;
• international trade;
• customs law;
• investment law;
• telecommunications regulation;
• data governance.

The more interconnected the digital economy becomes, the greater the potential conflict between:

territorial sovereignty

and

market integration.

XVIII. Competition-Law Analytical Framework

A regulator examining an AI border-control system could proceed through the following sequence:

Step 1 — Identify the relevant market

Determine whether the relevant market concerns:

• biometric verification;
• airport border technology;
• AI risk assessment;
• identity verification;
• cloud infrastructure;
• passenger-data processing.

Step 2 — Identify the geographic market

Ask whether the market is:

• national;
• regional;
• international;
• global.

Step 3 — Identify control points

Determine who controls:

• databases;
• APIs;
• algorithms;
• standards;
• hardware;
• authentication infrastructure.

Step 4 — Assess market power

Examine:

• market shares;
• entry barriers;
• network effects;
• switching costs;
• data advantages;
• regulatory certification.

Step 5 — Identify partitioning mechanism

Possible mechanisms include:

• geo-blocking;
• API restrictions;
• data localisation;
• exclusive licensing;
• interoperability restrictions;
• discriminatory access;
• algorithmic routing.

Step 6 — Assess foreclosure

Determine whether competitors are prevented from:

• entering;
• expanding;
• interoperating;
• accessing customers;
• accessing essential data.

Step 7 — Examine objective justification

Consider:

• security;
• privacy;
• cybersecurity;
• sovereignty;
• regulatory compliance.

Step 8 — Assess proportionality

Ask:

Could the legitimate objective be achieved through a less restrictive technical or commercial design?

XIX. Possible Competition Remedies

Authorities could consider:

1. Interoperability obligations

Require access to standardised APIs.

2. Non-discriminatory access

Require equivalent treatment of competing service providers.

3. Data portability

Permit lawful transfer of relevant data between systems.

4. Technical standards

Require interoperable technical specifications.

5. Separation remedies

Separate infrastructure operation from downstream commercial activities where necessary.

6. Transparency obligations

Require explanations of important access or ranking rules.

7. Non-exclusive procurement

Prevent unnecessarily exclusive long-term arrangements.

8. Monitoring trustees

An independent monitor could supervise compliance with access obligations.

9. Data-access safeguards

Permit access without compromising legitimate security or privacy requirements.

XX. Six Core Competition-Law Risks

XXI. Important Doctrinal Principle

The central lesson from the case law is that technology does not place territorial restrictions outside competition law merely because the restriction is implemented automatically.

The same competitive effect may arise through:

contract → software code → API → algorithm → database architecture

rather than through an explicit contractual prohibition.

Accordingly, modern competition authorities may need to examine not only contracts and correspondence but also:

• source-code functionality;
• API permissions;
• technical standards;
• access logs;
• algorithmic rules;
• data flows;
• model inputs;
• system architecture.

XXII. Conclusion

AI border-control systems sit at the intersection of national security, digital infrastructure, data governance, and competition law. Their territorial functionality can be legitimate because borders necessarily involve jurisdiction-specific rules. However, where a dominant undertaking uses border-control infrastructure, algorithms, data, APIs, or interoperability restrictions to create artificial commercial territories, traditional competition-law principles concerning market partitioning, refusal of access, discrimination, tying, self-preferencing, and leveraging become relevant.

The cases of Consten and Grundig, Football Association Premier League, GlaxoSmithKline, Pierre Fabre, Google Shopping, Slovak Telekom, Bronner, and Magill provide useful doctrinal foundations. They do not establish that a particular AI border-control arrangement is unlawful; rather, they supply principles for analysing whether technological territorialisation is genuinely regulatory or instead produces unjustified exclusion of competition.