Competition Law and Future Challenges of Competition Regulation in Autonomous Economies

1. Introduction

An autonomous economy is an economic environment in which increasingly sophisticated AI systems, autonomous agents, robots, algorithms, smart contracts and machine-to-machine platforms make or execute economic decisions with limited or no immediate human intervention.

Examples may include:

• AI agents autonomously purchasing goods and services;
• algorithmic firms autonomously setting prices;
• autonomous supply-chain systems selecting suppliers;
• AI agents negotiating contracts;
• machine-controlled financial trading;
• autonomous energy-management systems;
• AI platforms allocating computing resources;
• robots competing in manufacturing and logistics;
• autonomous marketplaces in which machines transact directly with one another; and
• decentralized systems using smart contracts to execute transactions automatically.

Traditional competition law generally assumes that human-controlled undertakings make strategic decisions. Autonomous economies challenge that assumption. The central future question becomes:

Who should bear competition-law responsibility when the economically significant decision is made by an autonomous machine rather than directly by a human decision-maker?

Existing competition law can address many of these problems through concepts such as undertaking, agreement, concerted practice, dominance, exclusionary conduct, merger control and abuse of market power. However, autonomous markets create significant problems concerning attribution, intent, algorithmic collusion, market definition, causation, transparency, remedies and enforcement.

The UK's Competition and Markets Authority has already identified access to inputs, diversity, choice, transparency, accountability and contestability as important principles for AI foundation-model markets.

2. Meaning and Characteristics of Autonomous Economies

An autonomous economy differs from an ordinary digital economy because the machine is not merely a tool used by a human decision-maker.

The machine may:

1. observe market conditions;
2. collect information;
3. predict demand;
4. select strategies;
5. negotiate;
6. determine prices;
7. enter transactions;
8. modify contractual terms;
9. switch suppliers;
10. respond to competitors; and
11. continuously optimize its conduct.

Thus, the traditional chain

Human → Decision → Algorithm → Market

may become:

Data → AI System → Autonomous Decision → Transaction → Market Feedback → New AI Decision

The competitive consequences can therefore emerge without a conventional human agreement.

3. Existing Competition Law Framework

Autonomous economies will primarily interact with the traditional pillars of competition law.

A. Anti-competitive agreements

The major question will be whether autonomous systems can produce an unlawful:

• agreement;
• concerted practice;
• coordination;
• information exchange;
• price-fixing mechanism; or
• market allocation.

B. Abuse of dominance

Autonomous platforms may acquire significant market power through:

• data accumulation;
• network effects;
• computational scale;
• control of AI infrastructure;
• control over foundation models;
• interoperability restrictions;
• exclusive access to data;
• self-preferencing;
• tying;
• refusal of access; and
• algorithmic discrimination.

C. Merger control

Future mergers may involve acquisitions of:

• AI models;
• autonomous-agent developers;
• data providers;
• compute providers;
• robotics companies;
• autonomous marketplaces;
• algorithmic infrastructure; or
• critical APIs.

A transaction may be competitively significant even where the acquired company has relatively little current revenue.

D. Consumer and business dependency

Autonomous agents may make users increasingly dependent on a small number of AI intermediaries. The competition issue therefore moves beyond price toward:

• choice;
• interoperability;
• innovation;
• access;
• quality;
• privacy;
• data portability;
• transparency; and
• technological neutrality.

4. Major Future Challenges

I. Attribution of Conduct to an Autonomous Agent

This may become the fundamental legal problem.

Suppose an autonomous purchasing agent independently decides to boycott Supplier X.

Who is responsible?

Possible candidates include:

• the developer;
• the owner;
• the operator;
• the deploying enterprise;
• the platform;
• the person who supplied the objective function; or
• the AI system itself.

Competition law traditionally regulates undertakings, not machines as independent legal persons.

Therefore, future legislation may need clearer attribution rules.

A possible principle could be:

Autonomous decision-making should not automatically eliminate responsibility where the relevant undertaking designed, deployed, controlled or economically benefited from the system.

5. Algorithmic and Autonomous Collusion

This is one of the most important challenges.

Traditional cartel law generally looks for:

Communication → Agreement → Coordinated Conduct

Autonomous systems may instead produce:

Observation → Prediction → Adaptation → Parallel Conduct

For example, five autonomous pricing systems could independently learn that maintaining a particular price maximizes profits.

No employee communicates with another employee.

No explicit cartel agreement exists.

Nevertheless, the economic result may resemble price coordination.

Legal difficulty

Competition authorities must distinguish between:

• legitimate independent adaptation;
• conscious parallelism;
• algorithmic coordination;
• tacit collusion;
• explicit collusion encoded into software; and
• coordination caused by a common third-party algorithm.

This distinction will become increasingly important.

6. The "Algorithm as Competitor" Problem

Autonomous systems may become genuine economic decision-makers.

Imagine two AI companies deploy autonomous agents that continuously:

• negotiate prices;
• buy inventory;
• select advertising space;
• negotiate transportation;
• purchase electricity; and
• enter derivatives transactions.

The agents may effectively become the strategic actors in the market.

Competition law will therefore need to determine whether the AI is merely:

an instrument of the undertaking

or whether it should be regarded as:

an economically autonomous decision-making mechanism whose conduct requires a special regulatory framework.

Existing doctrine generally points toward responsibility remaining with the undertaking rather than recognizing AI as an independent undertaking.

7. Autonomous Price Setting

Autonomous pricing creates several competition concerns.

AI systems can:

• change prices within milliseconds;
• discriminate among consumers;
• predict competitors' responses;
• identify price-sensitive customers;
• coordinate through market signals;
• dynamically restrict supply; and
• optimize prices across multiple markets.

Example

Suppose autonomous systems belonging to several competing airlines learn that reducing capacity simultaneously produces higher profits.

Even without explicit communication, the resulting conduct could generate substantial competition concerns.

Authorities will therefore require sophisticated economic and technical evidence.

8. Personalized Pricing and Algorithmic Discrimination

Autonomous agents may possess extensive information about individual consumers.

A system could theoretically determine:

Consumer A will pay ₹1,000; Consumer B will pay ₹1,800.

The competition problem is not necessarily price discrimination itself. The important questions are:

• Does the undertaking possess market power?
• Does discrimination exclude rivals?
• Does it exploit locked-in consumers?
• Does it reduce contestability?
• Does it facilitate coordination?
• Does it prevent entry?

Autonomous economies may therefore make traditional market-wide price analysis increasingly inadequate.

9. Autonomous Self-Preferencing

A dominant AI platform may operate:

1. a general AI assistant;
2. an AI marketplace;
3. a payment system;
4. an autonomous shopping agent; and
5. its own competing products.

The system could automatically recommend the platform's own products.

This creates a future version of self-preferencing.

The European Commission's Digital Markets Act enforcement has already addressed self-preferencing and interoperability issues involving major digital platforms. In July 2026, the Commission also adopted measures concerning AI interoperability on Android and access to Google Search data, illustrating how competition regulation is moving toward AI-related access and interoperability questions.

10. Data as a Competitive Asset

Autonomous economies will depend heavily on data.

Competitive advantages may arise from:

• proprietary datasets;
• real-time transaction data;
• behavioral data;
• industrial data;
• training datasets;
• search data;
• sensor data;
• transaction histories; and
• feedback generated by autonomous agents.

A dominant undertaking may obtain a data feedback loop:

More users → More data → Better AI → Better service → More users → More data

This can create substantial barriers to entry.

The CMA has specifically identified access to data, compute, expertise and funding as relevant to maintaining competitive AI markets.

11. Compute Infrastructure and Competition

Autonomous economies may depend upon:

• GPUs;
• cloud computing;
• specialized AI chips;
• model-serving infrastructure;
• energy;
• data centers;
• networking infrastructure.

Consequently, competition may shift from software markets toward compute infrastructure markets.

A firm controlling essential computational resources could potentially:

• discriminate against competitors;
• impose exclusive arrangements;
• bundle compute with AI services;
• restrict access;
• foreclose rival developers; or
• impose discriminatory technical conditions.

This creates a potential intersection between competition law and essential-facility/access doctrines.

12. Autonomous Marketplaces

Future marketplaces may contain almost no conventional human purchasing.

For example:

AI Buyer Agent → AI Seller Agent → AI Negotiator → Smart Contract → Automated Payment

This raises novel questions.

Who is the consumer?

The human ultimately benefiting from the transaction?

Who is the undertaking?

The AI developer?

The marketplace?

The owner of the AI agent?

What constitutes an agreement?

The machine's acceptance of another machine's offer?

What constitutes intent?

The objective function programmed by a developer?

These questions could require new statutory definitions.

13. Smart Contracts and Competition Law

Smart contracts can automatically execute transactions.

Once activated, they may:

• fix prices;
• restrict supply;
• allocate territories;
• impose loyalty conditions;
• execute rebates;
• exclude non-approved participants.

The difficulty is that the contractual restriction may execute automatically.

The absence of human intervention at the time of execution should not necessarily eliminate competition-law responsibility.

14. Autonomous Mergers and Acquisition Decisions

AI-driven companies may acquire competitors or complementary technologies at extremely rapid speeds.

Traditional merger control may be challenged by:

• acquisitions of low-revenue AI startups;
• acquisitions of talent;
• acquisition of datasets;
• acquisition of patents;
• acquisition of computing capacity;
• minority investments;
• serial acquisitions; and
• acquisitions intended to prevent future technological competition.

The concept of a killer acquisition may therefore expand from pharmaceuticals and digital platforms into autonomous AI ecosystems.

15. Market Definition in Autonomous Economies

Traditional market definition frequently asks:

What products or services are sufficiently substitutable?

In autonomous markets, substitution can become multidimensional.

For example, a consumer may choose between:

• a search engine;
• a chatbot;
• an AI personal assistant;
• an autonomous shopping agent;
• an integrated operating-system assistant.

The relevant market could therefore be difficult to identify.

Traditional SSNIP testing may also become less informative where services are:

• free;
• personalized;
• bundled;
• rapidly evolving; or
• provided through AI ecosystems.

Competition authorities may increasingly need to consider:

• functionality;
• data;
• attention;
• interoperability;
• compute;
• ecosystem dependency; and
• innovation competition.

16. Six Important Case Laws

The following cases do not all concern fully autonomous economies. They are important precedents by analogy because they establish principles that can be applied to algorithmic and autonomous markets.

1. United States v. Topkins, 2015

This is one of the most important algorithmic-pricing cases.

The US Department of Justice prosecuted sellers who used algorithms to coordinate prices for posters and other products sold online.

Principle

The use of an algorithm does not transform traditional price-fixing into lawful conduct.

Importance for autonomous economies

Future autonomous systems could similarly implement or facilitate cartel arrangements. Competition law can therefore focus on the underlying economic coordination rather than merely the technological mechanism.

2. United States v. Apple Inc., 791 F.3d 290 (2d Cir. 2015)

The case concerned Apple's role in the coordination of e-book pricing.

Principle

Technological innovation and sophisticated contractual arrangements do not immunize conduct from antitrust scrutiny.

Relevance

In an autonomous economy, companies cannot necessarily avoid competition liability merely because coordination is embedded in:

• software;
• digital contracts;
• APIs;
• AI systems; or
• automated pricing mechanisms.

3. Eturas UAB v. Lietuvos Respublikos konkurencijos taryba, C-74/14

The European Court of Justice considered an online booking platform through which a technical message communicated a maximum discount to travel agencies.

Principle

Participation in an electronically facilitated coordination mechanism can raise competition-law concerns even where communication occurs through an IT system rather than traditional personal meetings.

Relevance

This is highly significant for autonomous marketplaces.

An AI platform could theoretically become the mechanism through which competitors coordinate conduct.

The legal question would remain whether the participating undertakings knew or could reasonably be regarded as participating in the coordination.

4. AC-Treuhand AG v European Commission, C-194/14 P

The case concerned liability of an undertaking that facilitated cartel conduct without itself operating at the same level of the market.

Principle

Competition-law responsibility can extend to an entity that facilitates anti-competitive conduct.

Relevance

This principle may become important for:

• AI infrastructure providers;
• algorithm developers;
• autonomous marketplace operators;
• data intermediaries; and
• AI optimization providers.

A future issue will be whether an AI provider merely supplies neutral technology or knowingly facilitates anti-competitive coordination.

5. Google Shopping, Case AT.39740

The European Commission found that Google had abused a dominant position by giving preferential positioning to its comparison-shopping service.

Principle

A dominant platform may create competition problems by favoring its own downstream service.

Relevance to autonomous economies

An autonomous AI assistant could control:

• search;
• recommendations;
• purchasing;
• payment;
• logistics.

If the AI systematically favors affiliated businesses, autonomous decision-making could become a mechanism for algorithmic self-preferencing.

The later EU digital-platform regulatory framework has reinforced the importance of interoperability and non-discriminatory access.

6. Google Android, Case AT.40099

The European Commission's Android decision examined Google's use of contractual and technical restrictions affecting competing services.

Principle

Control over a major digital ecosystem can enable a dominant undertaking to extend market power into adjacent markets.

Relevance

Autonomous economies may be built around:

Operating System → AI Assistant → Marketplace → Payments → Logistics

If one undertaking controls the underlying ecosystem, it may be able to leverage that position into emerging autonomous-agent markets.

17. Additional Important Case Laws

7. Ohio v. American Express Co., 585 U.S. 529 (2018)

The US Supreme Court addressed two-sided transaction platforms and market definition.

Autonomous-economy relevance

AI marketplaces will frequently be two-sided or multi-sided:

• buyers;
• sellers;
• AI agents;
• developers;
• advertisers;
• payment providers.

Competition analysis must therefore consider interactions between different sides of the platform.

8. Qualcomm Inc. v. FTC, 969 F.3d 974 (9th Cir. 2020)

The case concerned competition issues surrounding technology licensing and alleged exclusionary conduct.

Relevance

Autonomous economies may depend on foundational technologies protected by:

• patents;
• standards;
• APIs;
• proprietary interfaces.

The case illustrates the difficulty of applying competition law to technology ecosystems where intellectual-property rights and market power overlap.

9. Intel Corp. v European Commission, C-413/14 P

The case concerned rebates and exclusionary effects.

Relevance

Autonomous systems may automatically offer:

• loyalty discounts;
• personalized rebates;
• algorithmically optimized incentives.

Future enforcement may therefore require sophisticated effects analysis rather than simply examining the existence of an automated pricing or rebate mechanism.

18. The Problem of Human Intent

Traditional competition law sometimes examines knowledge, intention or awareness.

Autonomous systems complicate this.

Suppose:

Developer A creates an AI whose objective is to maximize profit.

The AI discovers that refusing transactions with Supplier B increases profits.

Did Developer A intend to exclude Supplier B?

There are several possible approaches:

Approach 1 — Direct intent

Liability depends upon proof that humans intended the anti-competitive result.

Approach 2 — Reasonable foreseeability

Liability may arise where the undertaking could reasonably foresee that deployment of the system would produce anti-competitive effects.

Approach 3 — Risk-based responsibility

Operators of powerful autonomous systems may have continuing obligations to monitor and prevent foreseeable competition risks.

The third approach could become particularly relevant for highly autonomous systems.

19. Black-Box AI and Evidentiary Problems

Competition authorities traditionally rely on:

• emails;
• contracts;
• internal documents;
• meeting records;
• pricing records.

Autonomous systems may instead produce:

• billions of decisions;
• continuously changing models;
• neural-network parameters;
• dynamic objectives;
• machine-generated communications.

An authority may therefore know what happened without knowing why it happened.

This creates a major evidentiary problem.

Future competition investigations may require:

• algorithmic logs;
• model documentation;
• audit trails;
• training-data records;
• decision histories;
• model-version histories;
• API records;
• computational evidence; and
• reproducibility testing.

20. Autonomous Agents and Information Exchange

AI agents may continuously observe:

• competitors' prices;
• inventory;
• advertising;
• demand;
• capacity;
• consumer behavior.

This can produce a market in which competitors possess information that human firms could never obtain so quickly.

The danger is that real-time transparency can become anti-competitive transparency.

A perfectly transparent autonomous market may paradoxically become less competitive if every system immediately learns every competitor's strategy.

21. Autonomous Vertical Integration

A single AI company could operate:

Foundation Model → Cloud → Operating System → AI Agent → Marketplace → Payment → Logistics

This creates substantial vertical integration.

Potential concerns include:

• tying;
• bundling;
• foreclosure;
• self-preferencing;
• interoperability restrictions;
• discriminatory access;
• exclusive dealing;
• refusal to supply.

The challenge is determining when integration creates legitimate efficiency and when it suppresses competitive opportunities.

22. Interoperability as a Future Competition Remedy

Interoperability may become one of the most important remedies.

Possible obligations include requiring dominant AI systems to provide:

• API access;
• data portability;
• model interoperability;
• agent interoperability;
• identity portability;
• payment interoperability;
• search-data access;
• operating-system access.

Recent EU DMA measures concerning Android AI interoperability demonstrate how interoperability can be used to prevent platform control from restricting competing AI services.

23. Access to Data as an Antitrust Remedy

Competition authorities may increasingly consider:

Data access + interoperability + portability

as a combined competitive remedy.

For example, a dominant autonomous marketplace might be required to provide competitors with access to certain data under defined safeguards.

However, this raises conflicts with:

• privacy;
• cybersecurity;
• intellectual property;
• trade secrets;
• data protection;
• national security.

Therefore, data-sharing remedies must be carefully structured.

24. Competition Between AI Agents

An unusual future market may involve AI agents competing directly against one another.

For example:

Buyer Agent A

negotiates with

Seller Agent B

while

Logistics Agent C

negotiates with

Payment Agent D.

Human involvement may be limited to setting objectives.

Competition law may therefore have to regulate not only firms but also networks of autonomous economic agents.

25. Autonomous Economies and Essential Facilities

Certain infrastructure may become indispensable:

• cloud computing;
• AI accelerators;
• operating systems;
• model APIs;
• identity infrastructure;
• payment rails;
• autonomous mobility networks;
• energy-management networks.

Where a dominant undertaking controls an indispensable input, refusal or discriminatory access may become a competition concern.

The traditional essential-facilities doctrine may therefore acquire renewed importance in technologically autonomous markets.

26. Autonomous Agents and Consumer Lock-In

An AI agent may learn a user's:

• preferences;
• purchasing habits;
• financial arrangements;
• communication style;
• subscriptions;
• travel preferences;
• health-related choices.

Switching to another AI system may therefore become costly.

This creates a new form of behavioral and informational lock-in.

Competition remedies may consequently include:

• data portability;
• agent portability;
• interoperability;
• standardized formats;
• switching rights.

27. Competition Law and Autonomous Financial Markets

AI systems may independently:

• trade securities;
• rebalance portfolios;
• lend money;
• determine credit;
• manage liquidity;
• execute derivatives.

If multiple autonomous systems respond to the same signals, they could produce highly correlated conduct.

Potential competition risks include:

• algorithmic coordination;
• market manipulation;
• exclusion;
• discriminatory access;
• information advantages.

Competition law may increasingly intersect with financial-market regulation.

28. Autonomous Economies and Innovation Competition

Price may become less important where autonomous systems provide services at very low or zero monetary prices.

Competition may instead occur through:

• innovation;
• accuracy;
• speed;
• privacy;
• reliability;
• autonomy;
• interoperability;
• quality;
• safety.

Therefore, future antitrust analysis may have to give greater weight to innovation competition.

29. Future Regulatory Model

A future regulatory framework for autonomous economies could contain seven layers:

Layer 1 — Traditional Competition Law

• cartels;
• abuse of dominance;
• mergers;
• restrictive agreements.

Layer 2 — Algorithmic Accountability

• auditability;
• logging;
• explainability;
• monitoring.

Layer 3 — Interoperability

• APIs;
• data portability;
• agent interoperability.

Layer 4 — AI Infrastructure Access

• compute;
• cloud;
• data;
• model access.

Layer 5 — Algorithmic Collusion Controls

• detection;
• monitoring;
• testing;
• reporting.

Layer 6 — Autonomous-System Governance

• human oversight;
• accountability;
• risk assessment;
• compliance systems.

Layer 7 — International Cooperation

Autonomous digital markets operate across borders, requiring cooperation among:

• EU competition authorities;
• US antitrust agencies;
• UK CMA;
• Asian competition authorities;
• other national regulators.

The UK, EU and US competition authorities have already articulated common competition principles for generative AI foundation-model markets, indicating increasing international convergence around issues such as access, choice, innovation and contestability.

30. Possible Future Legal Tests

Future competition law may need to develop tests such as:

A. Autonomous Decision Test

Whether a decision was materially generated by an autonomous system.

B. Foreseeability Test

Whether the undertaking could reasonably foresee the competitive consequences.

C. Control Test

Who possessed effective control over the autonomous system?

D. Economic Benefit Test

Who benefited from the conduct?

E. Systemic Power Test

Whether the AI system controls an economically critical infrastructure or ecosystem.

F. Contestability Test

Whether competitors can realistically enter, interoperate and expand.

G. Auditability Test

Whether the undertaking can reconstruct significant autonomous decisions.

31. Key Compliance Duties for Future Autonomous Enterprises

Companies deploying autonomous economic systems should consider:

1. Competition-by-design
2. Algorithmic competition-impact assessments
3. Continuous cartel-risk monitoring
4. Independent algorithm audits
5. Model and decision logging
6. Human escalation procedures
7. Interoperability planning
8. Data-access compliance
9. Merger-control screening
10. Documentation of pricing algorithms
11. Monitoring of autonomous communications
12. Periodic competition-law stress testing

32. Core Legal Problems Summarized

33. Future Case-Law Direction

The existing cases suggest several principles that could shape future jurisprudence:

Topkins

Algorithms do not immunize cartel conduct.

Apple e-books

Digital technology does not eliminate traditional antitrust responsibility.

Eturas

Electronic platforms can facilitate competition-law violations.

AC-Treuhand

Facilitation can create competition-law exposure.

Google Shopping

Algorithmic ranking can become a mechanism of exclusionary self-preferencing.

Google Android

Control over a technological ecosystem can allow market power to be leveraged into adjacent markets.

Ohio v. American Express

Platform economics requires attention to multiple sides of the market.

Intel

Automated discounts and incentives may still require analysis of competitive effects.

34. Conclusion

The central challenge of competition regulation in autonomous economies is not simply that machines will make decisions faster. It is that the traditional legal concept of market conduct is built around identifiable human decisions, whereas autonomous economies may produce economically significant outcomes through continuously learning systems.

The most important future issues will therefore be:

1. attribution of autonomous conduct;
2. algorithmic and tacit coordination;
3. AI-driven pricing;
4. self-preferencing by autonomous platforms;
5. control of data and compute;
6. AI ecosystem dominance;
7. interoperability and portability;
8. autonomous marketplace governance;
9. evidence and algorithmic auditability;
10. AI acquisitions and emerging-market competition;
11. innovation competition; and
12. cross-border enforcement.

The existing case law demonstrates that competition law is capable of regulating technologically sophisticated conduct even when the mechanism is novel. The future challenge is to adapt established principles—agreement, undertaking, dominance, foreclosure, effects, market power and competitive harm—to circumstances where the economic actor may be an autonomous system operating continuously across interconnected markets.

In short, the future of competition regulation in autonomous economies is likely to move from regulating individual corporate decisions toward regulating the competitive architecture within which autonomous economic decisions are made.