Competition Law and Future-Proof Competition Frameworks for Machine-Mediated Economies

Introduction

A machine-mediated economy is an economy in which important commercial decisions are increasingly made, assisted, or implemented by software, algorithms, artificial intelligence (AI), automated agents, ranking systems, recommendation engines, pricing algorithms, smart contracts, and autonomous platforms.

Traditional competition law generally assumes that human firms make commercial decisions and that competition occurs through relatively identifiable markets, products, prices, and contractual arrangements. Machine-mediated markets challenge these assumptions because:

• algorithms can determine prices without direct human intervention;
• AI systems can autonomously optimize commercial strategies;
• platforms can simultaneously act as market operators and competitors;
• access to data can determine market power;
• algorithms can personalize prices and rankings at enormous scale;
• network effects can rapidly produce concentrated markets;
• AI models may become infrastructure upon which competing businesses depend;
• automated systems can make exclusionary conduct difficult to detect;
• machine-to-machine interaction can facilitate coordination without an explicit human agreement.

A future-proof competition framework therefore requires competition law to move from a purely firm-centred and transaction-centred model toward a framework capable of examining data, algorithms, infrastructure, interoperability, ecosystems, computation, and automated decision-making.

I. Meaning of a Future-Proof Competition Framework

A future-proof framework is not necessarily a completely new antitrust statute. Rather, it is a legal and institutional architecture capable of applying established competition principles to technological environments that may not yet exist.

Its principal objectives should be:

1. preservation of competitive market structures;
2. prevention of durable technological bottlenecks;
3. protection of contestability;
4. prevention of algorithmically reinforced dominance;
5. preservation of innovation;
6. interoperability and portability where necessary;
7. detection of machine-facilitated collusion;
8. accountability for autonomous commercial systems;
9. effective merger control over emerging technologies;
10. technologically informed remedies.

II. Why Machine-Mediated Markets Create New Competition Problems

1. Algorithmic pricing

Algorithms can change prices continuously according to:

• demand;
• competitor prices;
• inventory;
• consumer behaviour;
• location;
• time;
• purchasing history;
• predicted willingness to pay.

The difficulty is that conventional cartel law normally looks for communication or agreement between competitors.

In machine-mediated markets, competitors may potentially reach similar pricing outcomes through automated systems without a conventional human meeting or explicit agreement.

Competition-law concern

Authorities must distinguish between:

Legitimate algorithmic adaptation

and

algorithmically facilitated coordination.

The mere fact that two algorithms produce similar prices should not automatically establish an infringement. The legal inquiry must examine the design, information flows, contractual arrangements, instructions, communications, and competitive effects.

III. Algorithmic Collusion

Algorithmic systems create several possible forms of coordination.

A. Direct coordination

Competitors intentionally program their systems to follow a common pricing rule.

B. Facilitated coordination

A common third-party algorithm supplies competing firms with pricing recommendations.

C. Tacit algorithmic coordination

Independent algorithms learn that maintaining elevated prices is commercially advantageous.

D. Predictive coordination

Algorithms continuously observe competitors and rapidly react to their behaviour, potentially making deviations from coordinated outcomes easier to detect.

This creates an important future competition-law question:

Can autonomous machine behaviour constitute legally relevant coordination when no human expressly agreed to coordinate?

The answer will depend upon the applicable jurisdiction's statutory requirements for agreement, concerted practice, intent, knowledge, causation and effects.

IV. AI and Market Power

AI introduces a new set of potential sources of market power.

A firm may possess:

• enormous datasets;
• computing capacity;
• specialised chips;
• cloud infrastructure;
• foundation models;
• proprietary training data;
• distribution networks;
• developer ecosystems;
• user feedback loops;
• application programming interfaces;
• technical standards.

Consequently, future dominance analysis cannot rely exclusively upon traditional measures such as market share.

A technologically sophisticated assessment may need to consider:

Data + Compute + Model + Distribution + Network Effects + Switching Costs + Interoperability + Ecosystem Control.

V. Data as a Competitive Asset

Data may function as:

• an input;
• a competitive advantage;
• a source of prediction;
• a barrier to entry;
• an ecosystem connector;
• an advertising asset;
• a training resource for AI.

However, possession of data should not automatically be treated as dominance.

Competition authorities should investigate whether the relevant data are:

1. difficult to replicate;
2. competitively significant;
3. sufficiently current;
4. necessary for particular products;
5. obtainable through alternative means;
6. protected by network effects;
7. capable of being transferred or interoperated.

The Bundeskartellamt's Facebook/Meta proceeding illustrates how competition law can interact with data governance. The authority prohibited Meta from combining certain user data from different sources without voluntary consent, and the CJEU confirmed that competition authorities may take data-protection rules into account in their competition-law assessment. The German proceeding was ultimately concluded in 2024 following implementation measures.

VI. Algorithmic Self-Preferencing

One of the most important machine-mediated competition problems is algorithmic self-preferencing.

A platform may operate:

the infrastructure + the ranking algorithm + its own competing product.

For example, an online marketplace could use an algorithm to determine which products receive prominent placement while simultaneously selling its own products.

The problem is not simply that the platform owns a private-label product. The competition concern arises when control over an essential digital interface can be used to systematically disadvantage rivals.

VII. Case Law

1. Google Search (Shopping) — European Union

Google and Alphabet v European Commission, Case C-48/22 P (2024)

This is one of the most important authorities for machine-mediated markets.

The European Commission found that Google gave preferential placement to its own comparison-shopping service while demoting competing comparison-shopping services through its general search algorithms.

The CJEU dismissed Google's appeal in September 2024 and upheld the €2.4 billion fine.

Significance

The case demonstrates that:

• algorithmic ranking can constitute commercially significant conduct;
• control over a digital gateway can create leveraging opportunities;
• competition law can examine discriminatory treatment embedded in ranking systems;
• technological neutrality does not mean technological blindness.

Future application

The principle may become relevant to:

• AI search engines;
• AI-generated recommendations;
• chatbot answers;
• autonomous purchasing agents;
• app stores;
• digital marketplaces;
• travel platforms;
• AI model marketplaces.

2. Google Android — European Union

Google Android, Commission Decision, 2018

The European Commission found Google dominant in several markets associated with Android and found that contractual restrictions concerning Android devices contributed to strengthening Google's search dominance.

The Commission identified restrictions involving device manufacturers, pre-installation and distribution arrangements, and related contractual practices.

Competition principle

The case demonstrates the importance of ecosystem leveraging.

A company may possess power in one technological layer and use that position to strengthen another.

Future significance

The same principle may apply to:

AI operating system → AI assistant → search → applications → payments → advertising.

Thus, competition authorities may increasingly need to examine vertical technological ecosystems rather than isolated markets.

3. Amazon Marketplace — European Commission

The European Commission's Amazon Marketplace investigation concerned the use of non-public seller data and potential preferential treatment within Amazon's marketplace ecosystem.

Amazon offered commitments concerning, among other things:

• use of non-public seller data;
• Buy Box selection;
• Prime eligibility;
• logistics arrangements;
• treatment of competing offers. 

Competition principle

The case illustrates the dual-role problem:

A platform can simultaneously be the marketplace operator and a participant in the marketplace.

That creates a structural conflict because the platform may possess information about competitors that those competitors cannot obtain themselves.

Future AI application

An AI platform could potentially observe:

• prompts;
• customer demand;
• conversion rates;
• developer behaviour;
• emerging products;
• commercially sensitive information.

If the platform uses that information to improve its own competing products, competition concerns may arise.

4. Meta/Facebook — Bundeskartellamt and CJEU

Bundeskartellamt Facebook proceeding; CJEU, Meta Platforms and Others

The case concerned Meta's combination of user data from Facebook, Instagram, WhatsApp and third-party sources.

The CJEU held that a competition authority may take GDPR considerations into account when assessing whether conduct constitutes an abuse of dominance.

Importance for machine-mediated economies

This case demonstrates that competition law cannot always treat:

• privacy;
• data;
• consumer autonomy;
• platform design;
• market power

as completely separate subjects.

Future implication

For AI systems, the relevant question may become:

Can the accumulation and combination of personal and behavioural data itself reinforce a competitive advantage that competitors cannot realistically reproduce?

This does not mean that every privacy violation is an antitrust violation. Rather, it demonstrates the possibility of interaction between regulatory regimes.

5. FTC v Facebook/Meta — United States

The FTC's Facebook litigation alleges that Facebook maintained monopoly power through acquisitions and restrictions affecting developers' access to its platform and APIs. The FTC's case remains a significant U.S. digital-platform antitrust proceeding.

Competition principle

The case highlights:

• network effects;
• switching costs;
• API access;
• acquisitions of potential competitors;
• platform dependency;
• innovation competition.

Future significance

Machine-mediated markets make potential competition particularly important.

An emerging AI company may have:

• low present revenue;
• relatively few users;
• but strategically important technology.

Traditional turnover-based merger analysis may therefore miss transactions involving nascent competitors.

Future merger control may need to consider:

innovation capability + data assets + technical talent + model capability + future competitive potential.

6. Epic Games v Apple — United States

Epic Games, Inc. v Apple Inc., 559 F. Supp. 3d 898 (N.D. Cal. 2021), aff'd in part and rev'd in part, 67 F.4th 946 (9th Cir. 2023)

The litigation examined Apple's App Store structure, including restrictions surrounding distribution and in-app payments.

The case is important because it demonstrates the difficulty of applying conventional antitrust principles to a multi-sided digital ecosystem.

The later Epic Games v Google litigation has likewise focused on app-store distribution, billing and platform restrictions. The U.S. Department of Justice's materials identify the 2021 Apple decision and subsequent Google litigation as important authorities in this area.

Future significance

The principles are relevant to AI ecosystems where one company controls:

• model distribution;
• application access;
• developer tools;
• payment systems;
• cloud infrastructure;
• model APIs.

A future competition framework therefore needs to analyse ecosystem control, not merely individual products.

7. U.S. v Google — Search and Search Distribution

The U.S. Department of Justice's Google search monopolization litigation provides another important modern example.

The district court found Google liable for monopolization involving general search and search advertising, while subsequent remedies have addressed distribution arrangements and access to search-related data and services.

Competition principle

The case demonstrates the significance of:

• default settings;
• distribution agreements;
• network effects;
• scale;
• access points;
• accumulated behavioural data.

Future AI significance

AI assistants may become the new gateway to information and transactions.

If an AI assistant becomes the default mechanism through which consumers:

• search;
• purchase;
• book travel;
• select financial products;
• access applications;
• communicate with businesses,

control over that assistant could become economically comparable to control over other critical digital gateways.

8. U.S. v Google — Digital Advertising Technology

The U.S. government's separate advertising-technology litigation resulted in a 2025 district-court finding that Google unlawfully monopolized parts of open-web digital advertising markets. The case subsequently moved into remedies proceedings, including further relief in 2026.

Importance

The case illustrates the competition risks arising when one firm operates multiple layers of an interconnected technological supply chain.

For example:

publisher → ad exchange → ad server → advertiser

can become an ecosystem where the same firm operates multiple layers.

AI analogy

A future AI ecosystem might similarly consist of:

chips → cloud → foundation model → AI agent → application marketplace → payment → advertising.

Competition authorities therefore need to consider vertical foreclosure across technological layers.

IX. Essential Facilities and AI Infrastructure

The traditional essential-facilities concept may become relevant to machine-mediated economies, although its application remains jurisdiction-specific and generally demanding.

Potential technological bottlenecks include:

• cloud computing;
• high-performance computing;
• specialised AI chips;
• critical datasets;
• interoperability interfaces;
• payment rails;
• app stores;
• digital identity systems;
• technical standards.

However, being technologically important does not automatically make an asset an essential facility.

Authorities would need to establish the relevant legal requirements concerning:

• indispensability;
• duplication;
• exclusionary conduct;
• feasibility of access;
• competitive harm.

X. Interoperability as a Future Competition Remedy

Interoperability may become one of the central tools of future antitrust.

Potential forms include:

1. API interoperability

Competitors receive technically meaningful access to APIs.

2. Data portability

Users can transfer relevant information between services.

3. Protocol interoperability

Competing services can communicate with one another.

4. Model interoperability

AI applications can switch between models.

5. Identity interoperability

Users can authenticate across competing services.

6. Payment interoperability

Users can choose alternative payment mechanisms.

Interoperability can reduce:

switching costs + lock-in + network effects.

XI. Machine-Mediated Merger Control

Traditional merger control often concentrates on:

• turnover;
• assets;
• market shares;
• overlaps.

These measures can be inadequate for technology companies whose present revenue is low but whose technology is strategically significant.

A future-proof framework should examine:

A. Data acquisition

Does the transaction combine uniquely valuable datasets?

B. Innovation competition

Does the transaction remove a future technological competitor?

C. AI capability

Does the acquisition eliminate an emerging AI technology?

D. Infrastructure control

Does the merger combine critical cloud, compute or distribution assets?

E. Ecosystem expansion

Does the transaction enable a dominant ecosystem to enter an adjacent market?

F. Talent acquisition

Could the transaction function effectively as an acquisition of technological capability rather than conventional assets?

XII. Algorithmic Personalised Pricing

Machine-mediated economies may permit firms to charge different consumers different prices.

Potential competition concerns include:

• discrimination;
• exploitation of information asymmetries;
• coordinated pricing;
• exclusionary pricing;
• targeted discounts;
• personalised foreclosure.

Competition law should distinguish between:

price differentiation that intensifies competition

and

algorithmic discrimination that weakens competitive constraints.

The relevant evidence may include:

• algorithmic instructions;
• training data;
• pricing outputs;
• testing records;
• internal objectives;
• competitor information;
• model architecture;
• audit logs.

XIII. Algorithmic Transparency

A future competition authority may require firms to preserve:

• model versions;
• training-data provenance;
• pricing logs;
• ranking changes;
• recommendation rules;
• API calls;
• automated decisions;
• human overrides;
• model evaluations.

This is important because conventional discovery may be insufficient when the relevant decision was produced by a complex machine-learning system.

XIV. Competition Law and Explainability

Complete disclosure of source code should not automatically become the default remedy.

It could:

• reveal trade secrets;
• facilitate gaming;
• compromise security;
• expose intellectual property;
• make regulatory compliance unnecessarily burdensome.

A proportional framework could instead require:

1. regulatory access;
2. independent technical audits;
3. controlled testing;
4. outcome monitoring;
5. documentation;
6. explainability concerning specific competitive decisions.

Thus:

Regulatory explainability need not equal complete public disclosure of algorithms.

XV. New Market-Definition Techniques

Traditional market definition remains relevant but may need technological adaptation.

Authorities may examine:

• zero-price services;
• attention;
• data;
• interoperability;
• multi-homing;
• switching costs;
• ecosystem dependency;
• quality;
• innovation;
• latency;
• computational capacity.

For AI services, competition may occur between:

• AI models;
• search engines;
• assistants;
• software ecosystems;
• cloud services;
• applications;
• human services.

The relevant competitive constraint may therefore be difficult to capture through a simple price-based SSNIP test.

XVI. Competition and Autonomous Agents

A particularly important future issue is the emergence of AI agents that negotiate directly with other AI agents.

For example:

Consumer AI agent → searches suppliers → negotiates price → selects product → completes transaction.

This changes the traditional buyer-seller relationship.

Potential issues include:

• automated supplier exclusion;
• discriminatory recommendations;
• collective pricing;
• manipulation of consumer agents;
• preferential access;
• agent interoperability;
• autonomous purchasing agreements.

Competition law will therefore need to regulate not merely firms and platforms, but potentially the architectures through which commercial decisions are made.

XVII. Regulatory Sandboxes

Competition authorities could create controlled environments in which new technologies are tested before large-scale deployment.

A competition sandbox could examine:

• algorithmic pricing;
• AI recommendation systems;
• autonomous agents;
• interoperability;
• data-sharing arrangements;
• blockchain markets;
• smart contracts.

This would allow regulators to identify competition problems before network effects make intervention substantially more difficult.

XVIII. Continuous Competition Monitoring

Traditional antitrust is often retrospective:

Conduct occurs → investigation → decision → remedy.

Machine-mediated markets may require greater use of:

continuous monitoring → early warning → intervention → remedy → monitoring.

This could involve:

• algorithmic market surveillance;
• market concentration indicators;
• switching-cost measurements;
• interoperability testing;
• merger monitoring;
• automated anomaly detection.

The objective should be early identification of structural competitive deterioration without treating unusual algorithmic behaviour as automatically unlawful.

XIX. Dynamic Market Power

Market power in machine-mediated economies can change extremely rapidly.

A company could move from:

startup → platform → dominant ecosystem

within a comparatively short period.

Consequently, competition authorities should examine dynamic market power, including:

• speed of technological scaling;
• data accumulation;
• network effects;
• ecosystem expansion;
• user lock-in;
• learning effects;
• switching costs;
• access to compute.

Market shares should therefore be treated as evidence, rather than the sole measure of competitive power.

XX. Future-Proof Merger Remedies

Possible remedies include:

Structural remedies

• divestiture;
• separation of business units;
• ownership restrictions.

Behavioural remedies

• non-discrimination;
• interoperability;
• data-use restrictions;
• access obligations.

Technical remedies

• APIs;
• data portability;
• interoperability protocols;
• algorithmic auditing.

Governance remedies

• independent compliance monitors;
• technical trustees;
• periodic reporting;
• algorithmic-impact assessments.

Temporal remedies

Remedies could also contain sunset clauses and periodic reviews, because technological conditions change rapidly.

XXI. Proposed Future-Proof Competition Framework

A comprehensive framework can be represented as follows:

Layer 1 — Market Structure

Identify:

• market shares;
• network effects;
• switching costs;
• barriers to entry.

↓

Layer 2 — Digital Infrastructure

Examine:

• cloud;
• compute;
• APIs;
• app stores;
• payment systems;
• data infrastructure.

↓

Layer 3 — Algorithmic Conduct

Investigate:

• ranking;
• pricing;
• recommendations;
• self-preferencing;
• exclusion;
• coordination.

↓

Layer 4 — Data

Assess:

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

↓

Layer 5 — Innovation

Consider:

• nascent competitors;
• R&D;
• technological trajectories;
• potential competition.

↓

Layer 6 — Ecosystem Effects

Analyse:

• vertical integration;
• platform dependency;
• interoperability;
• ecosystem expansion.

↓

Layer 7 — Remedies

Select:

• behavioural;
• structural;
• interoperability;
• data;
• technical;
• monitoring remedies.

XXII. Core Principles of a Future-Proof Model

1. Technology neutrality

The law should regulate competitive harm rather than a particular technology.

2. Algorithmic neutrality

An algorithm should not receive immunity merely because the relevant decision was automated.

3. Innovation sensitivity

Intervention should account for both short-term consumer effects and long-term innovation competition.

4. Data neutrality

Data should neither automatically be regarded as an essential facility nor ignored as a potential source of durable market power.

5. Ecosystem analysis

Competition authorities should examine relationships among interconnected technological layers.

6. Contestability

Markets should remain open to new entrants rather than merely producing formally lawful but practically unchallengeable dominant ecosystems.

7. Proportionality

Regulation should not unnecessarily suppress legitimate technological innovation.

8. Auditability

Important automated commercial decisions should be capable of meaningful regulatory examination.

9. Interoperability

Where justified by competitive conditions, interoperability can reduce technological lock-in.

10. Institutional adaptability

Competition authorities need technical expertise, economists, data scientists, engineers and legal specialists working together.

XXIII. Six Major Doctrinal Lessons from the Case Law

XXIV. Major Challenges

A future-proof framework also creates risks for regulators.

1. False positives

Not every algorithmic similarity represents collusion.

2. Over-regulation

Excessive intervention may discourage technological innovation.

3. Technical complexity

Competition authorities may struggle to understand sophisticated AI systems.

4. Trade secrets

Regulatory transparency must coexist with legitimate intellectual-property protection.

5. Jurisdiction

Machine-mediated markets operate globally while competition laws remain predominantly territorial.

6. Regulatory fragmentation

Competition law, AI regulation, privacy law, consumer protection and cybersecurity law may overlap.

7. Rapid technological change

A remedy designed for today's technology may become obsolete tomorrow.

XXV. Conclusion

Competition law for machine-mediated economies must evolve from regulating only firms and transactions toward regulating competitive structures, technological ecosystems and automated decision-making processes.

The principal transformation is:

Traditional Competition Law → Digital Competition Law → Algorithmic Competition Law → Machine-Mediated Competition Governance.

The Google Shopping case demonstrates the importance of algorithmic ranking; Google Android demonstrates ecosystem leveraging; Amazon illustrates platform conflicts involving proprietary information; Meta demonstrates the interaction between data and competition; Epic Games illustrates multi-sided platform control; and modern Google litigation demonstrates the increasing importance of infrastructure, distribution and interconnected technological markets.

The future-proof model should therefore combine traditional antitrust doctrine with algorithmic auditing, dynamic market analysis, interoperability, data governance, technological merger review, continuous monitoring and adaptive remedies.

The central principle is not that machines should be regulated simply because they are machines. Rather, competition law should remain capable of identifying and remedying exclusionary, exploitative or coordinative conduct when commercial decision-making moves from humans to increasingly autonomous technological systems.