Competition Law and Competition Governance in Hyper-Automated Economies

Introduction

A hyper-automated economy is an economic environment in which a substantial part of production, distribution, pricing, procurement, logistics, marketing, investment, and consumer interaction is performed or coordinated by algorithms, artificial intelligence, autonomous agents, robotics, connected platforms, and machine-to-machine systems.

Competition law in such an economy faces a fundamental change: traditional competition law generally examines the conduct of identifiable human decision-makers and firms, whereas hyper-automation can produce competitive effects through automated decisions occurring at enormous speed, scale, frequency, and complexity.

The principal competition concerns include:

• algorithmic collusion;
• autonomous pricing;
• algorithmic discrimination;
• self-preferencing;
• platform foreclosure;
• data concentration;
• AI-driven exclusion;
• automated tying and bundling;
• interoperability restrictions;
• algorithmic mergers;
• exclusionary access rules;
• automated procurement;
• machine-to-machine coordination; and
• concentration of computational and infrastructural resources.

Competition governance therefore requires more than conventional ex-post enforcement. It increasingly requires continuous monitoring, algorithmic accountability, data-access rules, interoperability, transparency, merger scrutiny, and technologically informed remedies.

I. Meaning of Hyper-Automated Economies

A hyper-automated economy goes beyond ordinary digitalisation.

Traditional economy

Human → decision → transaction → market effect

Digital economy

Human + software → decision → transaction → market effect

Hyper-automated economy

AI/algorithm → autonomous decision → millions of transactions → continuous adaptation → market effect

Examples include:

1. AI-powered dynamic pricing;
2. autonomous supply-chain management;
3. algorithmic procurement;
4. autonomous financial trading;
5. robotic warehouses;
6. AI-controlled advertising;
7. automated credit allocation;
8. autonomous vehicles and logistics;
9. machine-to-machine contracting;
10. AI-based marketplace ranking.

The competition authority consequently has to examine not only what a company did, but also how its technological architecture produced the competitive outcome.

II. Competition Law Framework

The principal competition-law tools remain:

1. Prohibition of anti-competitive agreements

Automated systems may facilitate:

• price fixing;
• market allocation;
• output restrictions;
• information exchange;
• coordinated bidding;
• algorithmic signalling.

The absence of direct human communication does not necessarily eliminate the possibility of an infringement.

2. Abuse of dominance

A dominant technology platform may use automation to:

• favour its own services;
• restrict competitors' access;
• discriminate between trading partners;
• impose tying arrangements;
• exploit data advantages;
• degrade interoperability;
• restrict switching.

3. Merger control

AI and automation can make acquisitions particularly significant because a transaction may combine:

• datasets;
• computing infrastructure;
• algorithms;
• distribution platforms;
• AI models;
• cloud infrastructure;
• robotics systems;
• customer interfaces.

A transaction that appears small in conventional turnover terms may nevertheless eliminate an important future competitor.

4. Essential facilities and access

Hyper-automated markets may depend upon infrastructures such as:

• cloud computing;
• semiconductor capacity;
• AI models;
• datasets;
• payment networks;
• APIs;
• operating systems;
• automated logistics networks.

Competition authorities may therefore face questions concerning discriminatory or exclusionary access.

III. Major Competition Concerns

1. Algorithmic Collusion

Algorithms can observe competitors' prices and modify their own prices automatically.

A sufficiently sophisticated system could theoretically:

1. monitor competitors;
2. identify deviations;
3. predict reactions;
4. adjust prices;
5. punish aggressive discounting; and
6. maintain supra-competitive prices.

The legal difficulty is determining whether such conduct constitutes:

• an agreement;
• concerted practice;
• conscious parallelism;
• unilateral algorithmic conduct; or
• an abuse of dominance.

IV. Algorithmic Pricing and Tacit Coordination

Traditional markets often require human communication to establish a cartel.

Hyper-automated markets can potentially produce coordination through:

• common pricing algorithms;
• common software vendors;
• shared datasets;
• machine-readable market signals;
• algorithmic learning;
• automated responses.

This creates an important distinction between:

communication between firms and technological conditions facilitating coordination.

Competition governance must therefore examine the architecture through which coordination occurs.

V. Data as a Competitive Resource

Data can constitute an important competitive input.

A hyper-automated firm may possess:

• transaction data;
• behavioural data;
• location information;
• product data;
• search histories;
• supplier information;
• real-time market data.

Large datasets can improve AI systems, which produce better predictions, which generate more transactions, which generate more data.

This can create a data-feedback loop:

More users → more data → better AI → better service → more users → still more data.

Competition law therefore has to examine whether data accumulation creates:

• entry barriers;
• exclusionary advantages;
• network effects;
• economies of scale and scope;
• discriminatory access conditions.

VI. Self-Preferencing by Automated Systems

A vertically integrated platform may use algorithms to automatically favour its own products.

For example:

Platform → ranking algorithm → own product → greater visibility → increased sales → more data

The concern becomes stronger where the platform controls both:

1. the infrastructure through which competitors reach consumers; and
2. the competing service.

Automated self-preferencing may occur in:

• search results;
• online marketplaces;
• app stores;
• digital advertising;
• payment systems;
• travel platforms;
• food-delivery systems.

VII. Automated Tying and Bundling

Hyper-automated ecosystems may automatically combine products.

Examples:

• operating system + search;
• cloud + AI model;
• payment system + marketplace;
• hardware + software;
• app store + payment service;
• smart device + subscription.

The legal question is whether automation makes the bundled product more efficient or whether it forecloses competing suppliers.

VIII. Interoperability

Interoperability becomes particularly important where automated systems interact continuously.

Restrictions may include:

• API denial;
• incompatible protocols;
• technical barriers;
• restricted data portability;
• proprietary communication standards;
• closed AI ecosystems.

An incumbent can potentially make competitors less effective simply by preventing their systems from communicating with its infrastructure.

IX. Automated Procurement

AI can independently:

• identify suppliers;
• request quotations;
• compare bids;
• negotiate terms;
• select suppliers;
• execute contracts.

Competition concerns include:

• coordinated bidding;
• exclusionary procurement criteria;
• algorithmic supplier discrimination;
• automatic preference for incumbent suppliers;
• opaque tender evaluation.

Competition authorities may therefore need access to procurement-system records and decision logs.

X. Autonomous Agents as Market Participants

One of the most difficult questions is whether an AI agent should be treated merely as a tool of the firm.

At present, the central legal responsibility generally remains with the economic actor controlling the system rather than with the software itself.

Thus:

AI autonomy does not automatically create legal autonomy.

A company cannot necessarily avoid competition-law responsibility merely by arguing that its algorithm independently generated the conduct.

XI. Relevant Case Laws

1. United States v. Topkins

United States District Court, Northern District of California, 2015

Topkins involved an online poster market where sellers used algorithms in connection with an agreement to fix prices.

Importance

The case is highly relevant to hyper-automated markets because it demonstrates that:

• online commerce does not remove cartel liability;
• algorithms can be instruments of collusion;
• digital pricing systems can facilitate traditional cartel arrangements.

The central lesson is that automation does not immunise an underlying anti-competitive agreement.

2. Eturas v. Lietuvos Respublikos Konkurencijos Taryba

Court of Justice of the European Union, Case C-74/14, 2016

The case concerned an online booking system in which a common software system facilitated a restriction on discounts offered by travel agencies.

Importance

The Court considered circumstances in which information transmitted through a common electronic system could support an inference of participation in a concerted practice.

The case demonstrates that:

• electronic systems can facilitate coordination;
• firms cannot necessarily ignore restrictive information communicated through common software;
• competition law must account for technological communication mechanisms.

It is particularly relevant to platform-mediated coordination.

3. AC-Treuhand v European Commission

Court of Justice of the European Union, Case C-194/14 P, 2015

AC-Treuhand concerned an undertaking that facilitated cartel activity without itself operating as a traditional producer in the affected market.

Importance

The decision illustrates the broad approach to participation in anti-competitive arrangements.

For hyper-automated markets, the principle has relevance to:

• software providers;
• algorithm suppliers;
• technology intermediaries;
• automated coordination systems.

The fact that an actor does not itself sell the final product does not necessarily exclude competition-law responsibility where its conduct contributes to a cartel.

4. Google Shopping

European Commission, 2017

The European Commission found that Google had abused a dominant position by systematically favouring its comparison-shopping service in its general search results.

Importance

The case is significant for automated markets because ranking algorithms can determine:

• visibility;
• consumer access;
• traffic allocation;
• competitive opportunities.

It illustrates the competition-law importance of algorithmic ranking and self-preferencing.

5. Google Android

European Commission, 2018

The Commission found several practices concerning Google's Android ecosystem to be abusive, including restrictions relating to search and browser applications and licensing arrangements.

Importance

The case demonstrates how control over a technological ecosystem can produce competitive effects across connected markets.

Its relevance to hyper-automation includes:

• ecosystem power;
• tying;
• defaults;
• distribution restrictions;
• network effects;
• control over technological interfaces.

6. Google Search (AdSense)

European Commission, 2019

The European Commission addressed contractual restrictions concerning online search advertising intermediation.

Importance

The case demonstrates how contractual and technological restrictions within a digital ecosystem can affect competitors.

In hyper-automated advertising markets, algorithms determine:

• ad placement;
• bidding;
• matching;
• visibility;
• pricing.

Consequently, automated advertising infrastructure can become a major competition-law concern.

7. FTC v. Amazon

United States Federal Trade Commission and State Plaintiffs, 2023

The case concerns allegations relating to Amazon's conduct in online retail markets, including practices affecting sellers and competition.

Importance

The litigation illustrates modern competition concerns involving:

• marketplace power;
• seller relationships;
• pricing practices;
• self-preferencing;
• platform governance;
• control over marketplace infrastructure.

It demonstrates why competition authorities increasingly examine the architecture of digital platforms, rather than focusing exclusively on individual transactions.

8. European Commission v. Amazon — Marketplace Data

The European Commission's Amazon investigations concerning marketplace data addressed the use of non-public seller information by a vertically integrated platform.

Importance

The case highlights the competitive significance of data generated by third-party sellers.

In hyper-automated economies, platform data can be fed directly into:

• forecasting systems;
• pricing algorithms;
• inventory systems;
• product-selection systems;
• recommendation engines.

Consequently, control over data can translate into competitive advantages.

XII. Competition Governance Model

Hyper-automated economies require a layered governance model.

Layer 1 — Ex-ante regulation

Rules should address:

• interoperability;
• data portability;
• access;
• transparency;
• platform obligations;
• discriminatory algorithmic practices.

Layer 2 — Ex-post competition enforcement

Authorities should investigate:

• cartels;
• abuse of dominance;
• exclusion;
• discriminatory access;
• tying;
• predatory conduct.

Layer 3 — Algorithmic auditing

Authorities may need to examine:

• source-code documentation;
• model architecture;
• training data;
• decision logs;
• model updates;
• pricing histories;
• API interactions.

Layer 4 — Merger surveillance

Authorities should examine:

• acquisitions of AI startups;
• acquisition of datasets;
• cloud/AI combinations;
• vertical integration;
• killer acquisitions;
• ecosystem consolidation.

Layer 5 — Regulatory cooperation

Hyper-automated markets cross regulatory boundaries.

Competition authorities may therefore need cooperation with:

• data-protection authorities;
• financial regulators;
• telecommunications regulators;
• consumer-protection agencies;
• cybersecurity authorities;
• sectoral regulators.

XIII. Algorithmic Audit as a Competition Remedy

Where an algorithm is central to an infringement, conventional remedies may be insufficient.

Possible remedies include:

1. independent algorithmic audits;
2. monitoring trustees;
3. access obligations;
4. interoperability requirements;
5. data-sharing requirements;
6. algorithmic firewalls;
7. non-discrimination requirements;
8. prohibition of certain ranking practices;
9. deletion or separation of competitively sensitive datasets;
10. periodic compliance reporting.

The remedy must address the mechanism producing the competitive harm, not merely the visible outcome.

XIV. Competition and AI Explainability

A major problem is that some AI systems are difficult to interpret.

A competition authority may ask:

Why did the algorithm choose this price, supplier, ranking, or contractual condition?

If the undertaking cannot reconstruct the relevant decision, enforcement becomes difficult.

Accordingly, competition governance may increasingly require:

• audit trails;
• model documentation;
• version control;
• explainability mechanisms;
• retention of decision logs;
• reproducibility of important automated decisions.

XV. Human Oversight

Hyper-automation does not necessarily mean the elimination of human responsibility.

A governance system can maintain:

Human supervision → automated execution → monitoring → intervention → audit

This creates an important principle:

Automation may change the method of decision-making without eliminating the firm's legal responsibility for the competitive consequences of its system.

XVI. Competition Governance and Consumer Choice

Competition law ultimately protects the competitive process.

In hyper-automated markets, consumer choice may be influenced by:

• recommendation algorithms;
• personalised rankings;
• dynamic prices;
• automated discounts;
• behavioural targeting;
• default settings.

The consumer may therefore see only the output of a complex automated system rather than the competitive alternatives that existed behind it.

Competition governance must consequently examine whether automation:

• improves consumer choice;
• reduces search costs;
• increases innovation; or
• systematically limits exposure to competing alternatives.

XVII. Innovation Effects

Automation can generate substantial efficiencies.

It can:

• reduce transaction costs;
• improve logistics;
• reduce waste;
• increase production;
• improve forecasting;
• facilitate new products;
• lower search costs.

Competition law therefore should not treat automation itself as anti-competitive.

The relevant question is whether the deployment or control of automation materially restricts the competitive process.

This distinction is critical.

XVIII. Enforcement Challenges

1. Speed

Algorithms can make thousands of decisions before regulators can intervene.

2. Complexity

Modern AI systems can involve millions of parameters and multiple interacting models.

3. Attribution

It may be difficult to determine which firm, algorithm, developer, or platform generated the competitive effect.

4. Evidence

Traditional documentary evidence may be inadequate.

5. Cross-border operation

The algorithm, data, cloud infrastructure, company, and consumers may all be located in different jurisdictions.

6. Continuous modification

AI systems can change through:

• retraining;
• reinforcement learning;
• model updates;
• changing datasets.

Therefore, a system examined in January may behave differently several months later.

XIX. Proposed Competition-Governance Framework

A comprehensive framework can be represented as:

Automated Market

↓

Identify Market Power

↓

Identify Algorithmic Dependency

↓

Examine Data + Infrastructure

↓

Test Coordination / Exclusion

↓

Assess Consumer and Competitor Effects

↓

Examine Efficiency Justifications

↓

Algorithmic Audit

↓

Competition Remedy

↓

Continuous Monitoring

This represents a movement from static enforcement toward dynamic competition governance.

XX. Key Legal Principles

The principal principles emerging for hyper-automated economies are:

Principle 1 — Automation neutrality

The use of technology does not itself create an exemption from competition law.

Principle 2 — Accountability

A firm deploying an autonomous system should generally remain accountable for legally relevant conduct attributable to its business operations.

Principle 3 — Algorithmic transparency

Competition authorities should have sufficient access to understand technologically generated competitive effects.

Principle 4 — Data neutrality

Control of large datasets should not automatically be equated with dominance, but data advantages must be examined where they create significant competitive barriers.

Principle 5 — Interoperability

Control over critical technological interfaces may require competition scrutiny where it excludes competitors.

Principle 6 — Continuous enforcement

Rapidly changing markets may require ongoing monitoring rather than purely retrospective investigations.

Principle 7 — Innovation preservation

Competition governance should distinguish harmful exclusion from legitimate technological innovation and efficiency.

XXI. Comparative Significance of the Case Laws

XXII. Conclusion

Competition law in hyper-automated economies is evolving from a law of human transactions toward a law of technologically mediated market structures.

The central competition problems are no longer limited to traditional cartels and mergers. They increasingly involve:

• autonomous pricing;
• machine coordination;
• AI-driven exclusion;
• automated self-preferencing;
• data concentration;
• algorithmic discrimination;
• interoperability restrictions;
• AI ecosystem control;
• automated procurement; and
• technologically facilitated market foreclosure.

The most important governance challenge is therefore to maintain a balance between technological innovation and competitive market access.

The emerging model can be summarized as:

Competition law + algorithmic accountability + data governance + interoperability + merger scrutiny + continuous monitoring = competition governance for hyper-automated economies.