Competition Law and Future Governance of Interconnected Autonomous Markets

Introduction

Interconnected autonomous markets are markets in which pricing, contracting, allocation, recommendation, purchasing, logistics, financing, or other commercial decisions are increasingly made or assisted by AI systems, algorithms, autonomous agents, smart contracts, platforms, IoT devices, and interconnected digital infrastructure.

Competition law traditionally assumes identifiable firms making strategic decisions. Autonomous markets complicate this assumption because several economically significant decisions may be made simultaneously by systems that interact continuously with one another. A future competition-law framework must therefore address not only the conduct of individual firms but also the architecture, data flows, interoperability, algorithms, standards, and governance mechanisms through which autonomous systems compete.

The central competition-law questions include:

1. Can autonomous algorithms independently produce unlawful coordination?
2. Who is responsible when an AI system facilitates collusion?
3. Can interconnected platforms use data and interoperability advantages to exclude rivals?
4. When does an ecosystem become an essential competitive infrastructure?
5. How should competition authorities investigate algorithmic markets?
6. Can traditional concepts of dominance, market definition and relevant markets adequately capture autonomous ecosystems?
7. What remedies are appropriate where changing the algorithm itself is insufficient?

I. Meaning of Interconnected Autonomous Markets

An interconnected autonomous market contains several layers:

1. Autonomous decision-makers

AI systems may independently determine:

• prices;
• inventory;
• advertising bids;
• credit decisions;
• supply allocation;
• delivery routes;
• purchasing;
• investment decisions;
• customer targeting.

2. Interconnected platforms

Different platforms may exchange information through:

• APIs;
• cloud infrastructure;
• data-sharing arrangements;
• interoperability protocols;
• common standards;
• digital identity systems;
• payment networks.

3. Machine-to-machine interaction

Increasingly, software agents rather than human employees may negotiate with other software agents.

For example:

Supplier AI → marketplace AI → logistics AI → payment AI → consumer AI.

Each system may optimize its own objective while simultaneously affecting the competitive conditions faced by other participants.

4. Autonomous commercial ecosystems

The market may consequently become an ecosystem rather than a conventional bilateral transaction.

A single firm may control several interconnected layers:

Operating system → cloud → app marketplace → payment system → advertising → consumer data → AI model.

This creates competition concerns even where no individual contractual restriction appears obviously anticompetitive.

II. Traditional Competition Law and the Autonomous-Market Problem

Traditional competition law generally focuses on:

• agreements;
• concerted practices;
• unilateral conduct;
• abuse of dominance;
• mergers;
• exclusionary practices;
• exploitative conduct.

Autonomous markets introduce additional questions concerning:

III. Algorithmic Collusion

One of the most important future problems is algorithmic coordination.

Suppose several competing firms use autonomous pricing systems. Each algorithm observes competitors' prices and adjusts its own price.

No employee communicates with another competitor.

Nevertheless, the algorithms may converge toward a stable high-price outcome.

Competition-law issue

The key distinction is between:

A. Explicit coordination

Human actors deliberately instruct algorithms to coordinate.

B. Algorithm-facilitated coordination

Humans design algorithms knowing that they facilitate coordination.

C. Autonomous convergence

Algorithms independently learn strategies that result in coordinated outcomes.

The third category creates the greatest doctrinal difficulty.

Competition law cannot simply assume that an unlawful agreement exists whenever prices move together. Parallel conduct may have legitimate competitive explanations.

Future regulation may therefore require examination of:

• algorithm design;
• training objectives;
• available data;
• communication architecture;
• monitoring capabilities;
• algorithmic constraints;
• pricing instructions;
• expected competitor responses.

IV. Responsibility for Autonomous Conduct

A major future question is:

Who should bear responsibility for an autonomous algorithm's anticompetitive conduct?

Potentially responsible actors include:

1. the firm deploying the AI;
2. the developer;
3. the platform supplying the algorithm;
4. the data provider;
5. the cloud provider;
6. the intermediary coordinating the systems.

Competition law will generally need to preserve a principle of human and corporate accountability.

A company should not automatically avoid responsibility merely because an unlawful outcome was produced by an autonomous system.

At the same time, liability should not automatically be imposed on technology providers merely because their general-purpose technology was used by another company.

The critical question will increasingly be control, foreseeability, design, knowledge and economic function.

V. Dominance in Interconnected Autonomous Ecosystems

Market power may increasingly derive from control over an ecosystem rather than from conventional market share.

Important sources of power include:

1. Data advantages

Large datasets can improve:

• prediction;
• personalization;
• recommendation;
• fraud detection;
• pricing;
• AI training.

2. Network effects

More users generate more data, which improves the system, which attracts more users.

This creates a reinforcing cycle:

Users → Data → Better AI → More users → More data.

3. Switching costs

Users may become dependent on:

• accumulated data;
• digital identities;
• APIs;
• proprietary formats;
• software integrations;
• subscriptions.

4. Ecosystem leverage

A dominant company in one market may use its position to enter or control adjacent markets.

VI. Interoperability as a Competition Instrument

Future competition regulation may increasingly treat interoperability as a competitive principle.

A dominant platform could restrict competitors by refusing interoperability with:

• APIs;
• messaging systems;
• payment systems;
• cloud services;
• identity infrastructure;
• AI models;
• data portability mechanisms.

This creates a distinction between:

competition within an ecosystem and competition between ecosystems.

Where interoperability is technically feasible and competitively important, refusal to interoperate may potentially raise exclusionary-conduct concerns.

VII. Data as Competitive Infrastructure

Data may become comparable to other critical economic inputs.

Competition concerns can arise where a dominant undertaking:

• monopolizes commercially important datasets;
• prevents data portability;
• discriminates in data access;
• combines datasets across markets;
• uses exclusive data arrangements;
• denies rivals access to interoperability information.

However, competition law should distinguish between:

• genuinely indispensable data;
• data that competitors can reasonably reproduce;
• personal data subject to privacy restrictions;
• commercially confidential information;
• data that merely provides a competitive advantage.

VIII. Autonomous Agents and Digital Purchasing

Future consumers may use AI agents to purchase goods and services automatically.

For example:

Consumer AI → compares products → negotiates price → selects supplier → executes payment.

This could increase competition by reducing search costs.

But platforms could potentially manipulate autonomous agents through:

• preferential rankings;
• hidden commissions;
• exclusive access;
• personalized discrimination;
• interoperability restrictions;
• proprietary recommendation systems.

Competition authorities may therefore need to examine competition between AI agents as well as competition between the firms represented by those agents.

IX. Smart Contracts and Autonomous Agreements

Smart contracts can automatically execute transactions when predetermined conditions occur.

They may create efficiency by reducing:

• transaction costs;
• enforcement costs;
• administrative costs.

But automated contracts can also facilitate:

• resale-price restrictions;
• exclusionary clauses;
• automatic retaliation;
• coordinated pricing;
• discriminatory access.

A future regulatory system should therefore distinguish between automation of lawful competition and automation of unlawful restrictions.

X. Relevant Market Definition

Traditional market definition becomes harder where one ecosystem offers multiple interconnected services.

For example, a platform may simultaneously operate:

• search;
• advertising;
• payments;
• cloud computing;
• AI;
• marketplaces;
• logistics.

A narrow market-by-market approach may fail to capture competitive leverage.

Future analysis may therefore need to examine:

Horizontal relationships

Competition between equivalent services.

Vertical relationships

Competition between suppliers and downstream platforms.

Ecosystem relationships

Competition between entire technological ecosystems.

Data relationships

Control over inputs that improve multiple downstream products.

The relevant competitive constraint may consequently be dynamic and multidimensional.

XI. Key Case Laws

The following cases provide important foundations for understanding future governance of interconnected autonomous markets.

1. United States v. Apple Inc. — 2024

The U.S. Department of Justice brought an antitrust action against Apple concerning alleged exclusionary practices affecting smartphone competition.

The case illustrates the importance of:

• ecosystem control;
• interoperability;
• access restrictions;
• switching costs;
• control over complementary services;
• leveraging an established platform into adjacent markets.

Relevance

Future autonomous ecosystems may similarly generate competition concerns where one undertaking controls multiple technological layers and restricts rival access to those layers.

2. Ohio v. American Express Co. — 2018

The U.S. Supreme Court considered the competitive structure of the American Express transaction platform.

The Court emphasized the importance of understanding two-sided transaction markets as integrated platforms.

Relevance

Autonomous ecosystems may also connect multiple groups simultaneously:

Consumers ↔ AI agents ↔ merchants ↔ advertisers ↔ payment providers.

Competition analysis therefore cannot necessarily examine only one side of the ecosystem.

3. FTC v. Qualcomm Inc. — 2019

The case concerned Qualcomm's licensing practices and its position in cellular technology markets.

The litigation examined:

• intellectual property;
• licensing;
• technological standards;
• component markets;
• exclusionary effects.

Relevance

Interconnected autonomous markets may similarly depend upon technological standards and proprietary infrastructure.

Control over a critical technological layer can influence competition in downstream markets.

4. Google LLC v. Epic Games, Inc. — 2023

The litigation concerned Google's control of the Android application distribution ecosystem and alleged restrictions affecting app developers and alternative payment systems.

Relevance

The case illustrates how competition questions can arise from:

• app-store governance;
• payment restrictions;
• platform rules;
• developer access;
• ecosystem control.

These issues become even more important where autonomous agents increasingly choose and purchase digital services.

5. European Commission v. Google Shopping / Google Search (Shopping) — General Court, 2021

The European Union litigation concerning Google's comparison-shopping practices examined the use of dominance in general search to advantage another service.

The case is important for understanding:

• self-preferencing;
• platform leverage;
• search algorithms;
• ranking;
• discrimination against competitors.

Relevance

Autonomous markets may make self-preferencing more difficult to detect because rankings and recommendations may be continuously generated by AI.

Competition authorities may therefore need access to:

• ranking criteria;
• training information;
• audit logs;
• testing environments;
• relevant algorithmic documentation.

6. Google Android — European Commission / General Court

The European Commission's Android proceedings concerned Google's practices involving Android devices, application distribution and associated services.

The case illustrates competition concerns involving:

• tying;
• default arrangements;
• ecosystem control;
• mobile operating systems;
• foreclosure of competing services.

Relevance

The same principles can apply to future autonomous ecosystems in which one AI infrastructure controls access to multiple complementary markets.

7. Eturas — Case C-74/14

In Eturas, the Court of Justice of the European Union considered an online travel-booking system in which a platform transmitted a message concerning restrictions on discounts.

The case is particularly relevant to algorithmic coordination.

Relevance

It demonstrates that digital infrastructure can become an important mechanism through which competitors receive information capable of influencing their commercial behaviour.

Future cases may involve substantially more sophisticated versions of this problem, including automated pricing systems.

8. United States v. Airline Tariff Publishing Co.

The Airline Tariff Publishing litigation involved mechanisms through which airlines communicated pricing information through a computerized system.

Relevance

It is an important historical example of how technology can facilitate coordination among competitors.

Modern AI systems could perform similar information-processing functions at much greater speed and complexity.

9. T-Mobile Netherlands — Case C-8/08

The CJEU considered concerted practices involving competitors and emphasized that exchanges of strategically sensitive information can affect competitive behaviour.

Relevance

In autonomous markets, the same information-exchange problem may occur through:

• shared algorithms;
• common data providers;
• pricing software;
• industry platforms;
• machine-readable market information.

XII. Future Governance Model

A future framework could be structured around six governance layers.

Layer 1 — Conduct regulation

Continue applying:

• Article 101 TFEU;
• Article 102 TFEU;
• national competition laws;
• merger-control rules.

The objective would remain prevention of:

• cartels;
• exclusion;
• abuse of dominance;
• anticompetitive mergers.

Layer 2 — Algorithmic accountability

Require sufficiently powerful market participants to maintain:

• audit trails;
• algorithmic documentation;
• decision logs;
• risk assessments;
• governance records.

Layer 3 — Interoperability

Competition authorities could scrutinize unjustified restrictions on:

• APIs;
• data portability;
• technical standards;
• authentication;
• payment infrastructure.

Layer 4 — Data governance

Competition assessment should consider:

• access;
• portability;
• interoperability;
• exclusivity;
• data concentration.

Layer 5 — Merger control

Authorities may need to examine acquisitions involving:

• AI startups;
• datasets;
• cloud infrastructure;
• foundation models;
• autonomous-agent companies;
• interoperability technologies.

The concern is not simply the target's present revenue but potentially its future strategic significance.

Layer 6 — Regulatory cooperation

Future enforcement will increasingly require cooperation among:

• competition authorities;
• data-protection regulators;
• consumer-protection authorities;
• AI regulators;
• telecommunications regulators;
• financial regulators.

XIII. Ex Ante and Ex Post Regulation

A major future policy question concerns the appropriate balance between ex ante and ex post regulation.

Ex post approach

Authorities intervene after unlawful conduct occurs.

Advantages:

• preserves flexibility;
• avoids overregulation;
• allows markets to develop.

Problems:

• autonomous systems can evolve rapidly;
• harm may become entrenched before enforcement;
• network effects can make restoration difficult.

Ex ante approach

Certain powerful digital ecosystems may face obligations before misconduct occurs.

Potential requirements could include:

• interoperability;
• transparency;
• non-discrimination;
• data portability;
• restrictions on self-preferencing;
• merger notification.

A hybrid model is likely to be particularly relevant for highly interconnected markets.

XIV. Competition Remedies for Autonomous Markets

Traditional fines may be insufficient where the competitive problem is structural.

Potential remedies include:

1. Algorithmic separation

Separating competing commercial functions within a technological ecosystem.

2. Interoperability obligations

Requiring technically meaningful access.

3. Data portability

Allowing users or competitors to transfer relevant data under appropriate legal safeguards.

4. Non-discrimination obligations

Preventing a platform from systematically disadvantaging competing services.

5. Algorithmic audits

Independent examination of potentially problematic systems.

6. Structural remedies

In exceptional circumstances:

• divestiture;
• business separation;
• restrictions on acquisitions.

7. Merger conditions

Authorities could require:

• continued API access;
• licensing;
• interoperability;
• non-exclusive access;
• data separation.

XV. Role of Competition Authorities

Competition authorities of the future may need technological capabilities comparable to the firms they regulate.

They may require:

• AI expertise;
• data scientists;
• algorithm auditors;
• cybersecurity specialists;
• economists;
• competition lawyers;
• technical investigators.

Investigations may increasingly involve examination of:

Source code → model architecture → training data → prompts → system instructions → APIs → logs → outputs → market effects.

This does not mean competition authorities should regulate algorithms merely because they are complex. The focus should remain on competitive effects and legally relevant conduct.

XVI. Challenges of Proof

Autonomous markets create significant evidentiary problems.

Traditional evidence

• emails;
• contracts;
• meetings;
• telephone records;
• internal documents.

Autonomous-market evidence

• model weights;
• system prompts;
• API calls;
• telemetry;
• logs;
• training datasets;
• automated decisions;
• reinforcement-learning records;
• system updates.

Competition investigations may therefore increasingly require machine-readable evidence.

XVII. The Problem of Explainability

A company may argue:

"The AI made the decision independently."

That cannot automatically resolve competition-law responsibility.

A future legal framework may ask:

1. Who designed the objective?
2. What constraints were imposed?
3. What data was supplied?
4. What conduct was foreseeable?
5. Was the system monitored?
6. Were warnings generated?
7. Did the company modify the system after discovering the problem?

Thus, autonomy should not become a legal shield against accountability.

XVIII. Autonomous Mergers and Acquisitions

Future merger control may increasingly focus on acquisitions that appear small today but are strategically significant tomorrow.

Examples include acquisition of:

• foundation-model companies;
• autonomous-agent developers;
• specialized datasets;
• AI infrastructure;
• cloud tools;
• interoperability technologies;
• algorithmic marketplaces.

The traditional turnover-based approach may fail to identify certain strategically important transactions.

Authorities may therefore need stronger mechanisms for identifying nascent competition and innovation competition.

XIX. Future Governance Principles

A coherent competition regime for interconnected autonomous markets could be built around the following principles:

1. Human accountability

Autonomous technology should not eliminate corporate responsibility.

2. Competitive neutrality

Comparable firms should have meaningful opportunities to compete.

3. Interoperability

Critical ecosystems should not unnecessarily isolate users and rivals.

4. Data contestability

Control over data should not automatically become permanent competitive advantage.

5. Algorithmic accountability

Important autonomous decisions should be sufficiently auditable.

6. Dynamic competition

Authorities should consider future innovation and ecosystem evolution.

7. Proportionality

Intervention should correspond to demonstrated competitive harm.

8. Technological neutrality

Competition rules should regulate economic conduct rather than particular technologies merely because they are novel.

XX. Conceptual Framework

The future relationship can be represented as:

Autonomous Technology
↓
Data + Algorithms
↓
Interconnected Platforms
↓
Network Effects
↓
Ecosystem Power
↓
Potential Coordination / Exclusion
↓
Competition Investigation
↓
Algorithmic + Economic Analysis
↓
Behavioural / Structural Remedy
↓
Continuous Monitoring

This represents a shift from a firm-centric model toward an ecosystem-centric competition model.

XXI. Major Future Competition Risks

The principal risks are likely to include:

1. Algorithmic collusion
2. AI-enabled price discrimination
3. Self-preferencing by autonomous recommendation systems
4. Data monopolization
5. API exclusion
6. Interoperability restrictions
7. AI-agent manipulation
8. Algorithmic tying
9. Predatory automated pricing
10. Exclusive data arrangements
11. Anticompetitive standards
12. Killer acquisitions
13. Cross-market ecosystem leveraging
14. Automated discriminatory access
15. Common algorithmic infrastructure facilitating coordination

XXII. Conclusion

Competition law for interconnected autonomous markets will increasingly have to move beyond the traditional question:

"What did the firm do?"

toward a broader set of questions:

What system did the firm design?
What information did the system receive?
How did it interact with competing systems?
What competitive constraints existed?
Could rivals meaningfully interoperate?
Did the ecosystem allow competition to remain contestable?

The cases involving Apple, American Express, Qualcomm, Epic Games, Google Shopping, Google Android, Eturas, Airline Tariff Publishing and T-Mobile Netherlands demonstrate different components of this emerging framework: platform power, two-sided markets, technological infrastructure, tying, self-preferencing, information exchange and algorithmically facilitated coordination.

The future of competition law is therefore unlikely to be simply about regulating AI or autonomous systems as technologies. Its more fundamental task will be to ensure that autonomous economic systems remain contestable, interoperable, non-collusive and open to innovation, while preserving the flexibility necessary for legitimate technological development.