Competition Law and Competition Implications of Autonomous Legal Systems

1. Introduction

Autonomous Legal Systems (ALS) are AI-enabled systems capable of independently performing legal functions such as legal research, case analysis, contract review, drafting, compliance monitoring, litigation analytics, legal-risk assessment, negotiation support, and, in more advanced forms, taking sequential legal actions with limited human intervention.

Examples include AI legal-research platforms, autonomous contract-management systems, AI dispute-resolution systems, automated compliance engines, legal-document agents, and AI systems that interact with legal databases and enterprise workflows.

From a competition-law perspective, autonomous legal systems create a distinctive combination of:

• data concentration;
• control over legal databases and case-law repositories;
• AI-model concentration;
• vertical integration between legal content and AI tools;
• platform/network effects;
• algorithmic exclusion and self-preferencing;
• tying and bundling;
• interoperability and API-access problems;
• exclusive licensing of legal data;
• acquisitions of emerging legal-AI competitors; and
• algorithmically facilitated coordination among legal-service providers.

The Competition Act, 2002 in India addresses anti-competitive agreements, abuse of dominant position and combinations that cause or are likely to cause an appreciable adverse effect on competition.

The subject is particularly important because AI-powered legal research is already becoming integrated with established legal-information platforms. Westlaw, for example, now provides AI-assisted research grounded in its legal database, while Thomson Reuters describes its newer CoCounsel Legal product as an agentic AI system capable of research, verification and legal-work-product workflows.

2. Meaning of Autonomous Legal Systems

An autonomous legal system can be understood as:

A computational system that uses AI, legal databases, rules, models and automated decision processes to perform one or more legal functions with limited continuous human intervention.

There are several levels.

Level 1 — Assistive

The system merely assists a lawyer.

Examples:

• case-law search;
• document summarisation;
• citation checking;
• contract comparison.

Level 2 — Semi-autonomous

The system performs a legal task after receiving an instruction.

Examples:

• preparing a first draft of a pleading;
• identifying contractual risks;
• generating a legal memorandum;
• classifying regulatory obligations.

Level 3 — Agentic

The system independently performs multiple sequential activities.

For example:

1. receives a legal question;
2. searches legal databases;
3. identifies authorities;
4. analyses conflicting precedents;
5. drafts an opinion;
6. verifies citations;
7. prepares recommendations; and
8. updates the analysis when new law becomes available.

Level 4 — Highly autonomous legal infrastructure

The system may interact with:

• courts;
• regulators;
• contract platforms;
• corporate databases;
• compliance systems;
• arbitration platforms;
• legal-research databases; and
• other AI systems.

This level creates the most significant competition concerns because the autonomous system can become a gatekeeper between consumers and the legal-services ecosystem.

3. Relevant Markets

Competition authorities would first need to determine the relevant product and geographic markets.

Possible product markets include:

A. AI legal research

AI-powered systems that identify and analyse:

• cases;
• legislation;
• regulations;
• legal commentary;
• precedents.

B. Traditional legal databases

Examples include databases containing:

• judgments;
• statutes;
• regulations;
• annotations;
• headnotes;
• legal commentary.

C. AI legal drafting

Systems that generate:

• pleadings;
• contracts;
• legal opinions;
• compliance documents.

D. AI contract intelligence

Systems providing:

• contract review;
• clause extraction;
• risk scoring;
• automated negotiation.

E. Legal workflow platforms

Integrated systems combining:

• research;
• drafting;
• document management;
• litigation analytics;
• billing;
• compliance.

F. Underlying legal-data markets

A particularly important market may consist of access to:

• historical case law;
• annotated decisions;
• proprietary headnotes;
• citation networks;
• legal taxonomies;
• regulatory datasets.

The distinction is important because a company might not dominate the AI model itself but could possess significant market power because it controls a uniquely valuable legal-data resource.

4. Competition Law Issues

I. Dominance Through Legal Data

Legal information has special characteristics.

A large legal database may contain:

• decades of judgments;
• proprietary classifications;
• editorial annotations;
• citation relationships;
• headnotes;
• case histories;
• statutory relationships.

Once this information is incorporated into an AI system, the database can become an important competitive input.

The competition concern arises where a dominant legal-information provider:

1. controls an essential or highly valuable dataset;
2. provides access to its own AI product;
3. refuses or restricts access to competing AI developers; and
4. thereby strengthens its downstream position.

This resembles traditional essential-input/access problems.

The United States Thomson Reuters/Ross litigation illustrates the significance of this issue. Ross developed an AI legal-research product that competed with Westlaw, and disputes arose concerning access to Westlaw's proprietary legal content and the use of Westlaw-derived material in developing Ross's AI system. The litigation also contained antitrust counterclaims concerning Westlaw's market position.

5. II. Refusal to Supply Legal Data

A dominant legal-data provider could theoretically refuse access to:

• case-law APIs;
• citation databases;
• headnotes;
• legal metadata;
• historical datasets;
• machine-readable legislation.

Competition law may become relevant where access is indispensable for effective competition and refusal substantially forecloses downstream competitors.

However, not every refusal to license data is an abuse of dominance.

Authorities would generally examine:

• indispensability;
• availability of alternatives;
• duplication feasibility;
• investment incentives;
• justification for refusal;
• foreclosure effects;
• duration of exclusion.

This is closely connected with the principles developed in Bronner and IMS Health.

6. III. Tying and Bundling

An autonomous legal platform could bundle:

legal database + AI assistant + drafting tools + document management.

Suppose a dominant provider requires customers who want access to its case-law database to purchase its proprietary AI assistant.

Potential competition concerns include:

• foreclosure of competing AI products;
• increased switching costs;
• artificial expansion of market power;
• reduced innovation.

The Thomson Reuters v. Ross litigation is especially relevant because Ross alleged that Westlaw's public-law database was tied to Westlaw's search tools. The court analysed the need to establish separate products and define the relevant market before proceeding with the tying theory.

7. IV. Self-Preferencing

An autonomous legal ecosystem could favour its own products.

For example, an AI legal assistant integrated into a dominant legal database could systematically:

• recommend its own research service;
• rank its own legal templates first;
• direct users toward its affiliated law firms;
• privilege its own AI-generated authorities;
• suppress competing legal databases;
• make rival plugins less visible.

This creates a self-preferencing problem.

The classic modern competition-law reference is:

Google Shopping

The European Commission and EU courts considered Google's preferential treatment of its own comparison-shopping service within general search results.

The broader principle is relevant to autonomous legal systems because an AI platform can potentially control ranking and recommendation architecture rather than merely traditional search results.

8. V. Interoperability and API Restrictions

Autonomous legal systems require interoperability with:

• court databases;
• legal-research databases;
• document-management systems;
• contract-management platforms;
• e-discovery systems;
• government regulatory databases.

A dominant platform may make interoperability difficult by:

• restricting APIs;
• charging discriminatory access fees;
• limiting data portability;
• changing technical standards;
• throttling rival applications;
• denying access to important interfaces.

Competition law may therefore need to examine technical interoperability as a competitive parameter.

9. VI. Algorithmic Discrimination

Autonomous systems can discriminate against competing providers without an explicit human instruction.

For example, an AI platform might rank:

"Platform A — recommended"

while systematically assigning:

"Platform B — lower relevance"

even though both provide comparable services.

Such conduct may be difficult to detect because the discriminatory mechanism could be embedded in:

• model architecture;
• ranking algorithms;
• training data;
• reinforcement systems;
• recommendation systems;
• commercial objectives.

Competition authorities may therefore need to examine algorithmic outputs rather than only written corporate policies.

10. VII. Algorithmic Collusion

Autonomous legal systems could potentially be used by competing law firms or legal-service providers to:

• monitor competitors;
• adjust prices;
• respond automatically to competitor pricing;
• coordinate discounts;
• standardise fees.

The problem becomes particularly serious where autonomous systems learn from market information and independently adjust commercial strategies.

The traditional question is whether competition law requires:

human communication or agreement

when algorithms independently produce coordinated market outcomes.

This is one of the developing areas of AI competition law. Academic research already identifies AI-facilitated collusion, AI market power, AI exclusion and AI-related mergers as distinct competition-law problems.

11. VIII. Network Effects

Autonomous legal platforms can become stronger as more users participate.

More users produce:

• more queries;
• more feedback;
• more documents;
• more usage data;
• more workflow information.

This can improve the AI system, which attracts additional users.

The resulting cycle is:

Users → Data → Better AI → More Users → More Data

This creates significant barriers to entry.

A new competitor may possess an excellent AI model but still struggle because it lacks the enormous historical legal dataset accumulated by an incumbent.

12. IX. Switching Costs

Legal professionals may become deeply dependent upon one autonomous legal platform.

Switching costs can include:

• stored research histories;
• customised prompts;
• internal knowledge bases;
• firm-specific workflows;
• templates;
• citation structures;
• AI-agent configurations;
• integrations;
• client-matter databases.

Therefore, even if a competing system offers better technology, customers may remain locked into the incumbent.

Competition authorities may investigate whether contractual or technical restrictions unnecessarily increase those switching costs.

13. X. Exclusive Data Licensing

An incumbent AI legal platform could enter exclusive agreements with:

• courts;
• publishers;
• legal databases;
• regulatory-information providers;
• law firms;
• legal publishers.

If competing AI systems cannot obtain equivalent data, exclusivity may foreclose rivals.

The competitive analysis would depend upon:

• duration;
• scope;
• importance of the data;
• availability of alternatives;
• market coverage;
• foreclosure percentage.

14. XI. Mergers and Acquisitions

Autonomous legal systems create new merger-control concerns.

A large legal-data company might acquire:

• an AI legal startup;
• a legal workflow platform;
• an automated contract provider;
• a legal chatbot;
• an AI compliance company.

Even if the target has minimal current revenue, it may possess:

• important technology;
• unique training data;
• skilled AI researchers;
• strong user adoption;
• valuable patents;
• strategic interoperability.

Therefore, traditional turnover-based merger thresholds may fail to identify strategically important acquisitions.

The relevant theory resembles nascent-competition/acquisition concerns in digital markets.

15. XII. Killer Acquisitions

Suppose a dominant legal-information company acquires an emerging AI company.

The target may currently have:

• low revenue;
• few customers;
• substantial losses.

Nevertheless, it might become a significant competitive threat.

Competition authorities could therefore examine:

1. innovation potential;
2. technology quality;
3. user growth;
4. data assets;
5. intellectual property;
6. strategic importance;
7. likelihood of future competition.

16. XIII. Predatory Pricing

An incumbent autonomous legal platform might provide AI legal services:

free or substantially below cost

for an extended period.

If the objective or effect is to eliminate competing providers, competition authorities could consider predatory-pricing theories.

However, AI markets make cost measurement difficult because:

• marginal computational cost can be low;
• development costs are enormous;
• data acquisition costs may be sunk;
• cloud-computing costs fluctuate;
• services may initially be offered free to build network effects.

Traditional price-cost tests may therefore require adaptation.

17. XIV. Exclusive Integration

An autonomous legal platform might require:

"Use our AI only with our legal database."

Alternatively, a dominant document-management platform might prevent third-party AI systems from accessing its documents.

This can create:

• vertical foreclosure;
• exclusion of competitors;
• ecosystem lock-in;
• reduced interoperability.

The competition analysis would be especially important where the platform controls an unavoidable gateway.

18. XV. Professional-Legal Market Effects

Autonomous legal systems can affect competition not only among technology companies but also among law firms and lawyers.

Possible effects include:

Positive effects

• lower research costs;
• lower transaction costs;
• easier access to legal information;
• improved small-firm competitiveness;
• faster legal drafting;
• increased innovation;
• expanded access to legal services.

Potentially restrictive effects

• concentration of legal technology;
• dependence on a few AI providers;
• exclusion of smaller firms;
• increased data asymmetry;
• reduced technological diversity;
• automated discrimination between legal-service providers.

Thus, competition policy must distinguish between efficiency-enhancing automation and strategic exclusion.

19. Six Major Case Laws

Case 1 — United States v. Thomson Corporation, 949 F. Supp. 907 (D.D.C. 1996)

Facts

The case concerned the proposed acquisition of West Publishing by Thomson Corporation.

The government examined competition in online legal research, including products supplied to Lexis-Nexis.

The court recognised that Westlaw and Lexis-Nexis were major competitors in comprehensive online legal research and considered whether the transaction could reduce competition by limiting access to important Thomson products.

The case specifically identified concerns surrounding access to products such as Auto-Cite, which Lexis-Nexis used in competing with Westlaw.

Competition principle

The case demonstrates the importance of:

• legal-information databases;
• interoperability;
• access to complementary legal information;
• vertical relationships;
• merger effects in concentrated information markets.

Relevance to autonomous legal systems

An autonomous legal AI provider controlling a major legal database could become an important upstream supplier to rival AI platforms.

A merger could therefore create incentives to:

• deny data access;
• increase licensing costs;
• reduce interoperability;
• disadvantage rival AI systems.

Case 2 — Thomson Reuters Enterprise Centre GmbH v. ROSS Intelligence Inc.

Facts

ROSS developed an AI-powered legal research product intended to compete with Westlaw.

ROSS sought access to Westlaw content but did not receive a licence. It subsequently obtained "Bulk Memos" derived from lawyers' use of Westlaw headnotes and used them in developing its AI search system. Thomson Reuters sued ROSS.

Competition significance

Although the central litigation involved intellectual-property issues, the case is highly relevant to competition analysis because it demonstrates the strategic importance of:

• proprietary legal databases;
• headnotes;
• legal taxonomies;
• AI training material;
• access to legal information.

The antitrust counterclaims also raised issues concerning Westlaw's alleged market power and tying.

Relevance

It illustrates a central question for autonomous legal systems:

Can control over a proprietary legal-information ecosystem become a competitive bottleneck for AI entrants?

Case 3 — IMS Health GmbH & Co. OHG v. NDC Health GmbH & Co. KG

Principle

The Court of Justice of the European Union developed important limits concerning refusal to license intellectual-property rights.

The case established stringent conditions for treating refusal to license as an abuse of dominance.

Relevance to autonomous legal systems

Suppose a dominant legal-AI company owns:

• proprietary legal data;
• unique classification systems;
• indispensable interoperability technology.

A competitor seeking access cannot automatically claim an antitrust right to the data.

The IMS Health principles help distinguish:

legitimate intellectual-property protection

from

exceptional exclusionary conduct requiring competition-law intervention.

Case 4 — Bronner v. Mediaprint

Principle

The CJEU considered when refusal of access to an infrastructure controlled by a dominant undertaking could constitute abuse of dominance.

The decision imposed demanding conditions for applying the essential-facilities concept.

Relevance

For autonomous legal systems, the analogy could arise where a dominant company controls:

• an essential legal database;
• an indispensable API;
• an exclusive legal-information infrastructure.

A competition authority would need to determine whether:

1. the input is indispensable;
2. there is no realistic alternative;
3. duplication is practically/economically impossible; and
4. refusal risks eliminating effective competition.

Case 5 — Microsoft Corp. v. Commission

Principle

The Microsoft litigation is particularly relevant to interoperability and leveraging.

The European Commission addressed Microsoft's withholding of interoperability information and tying of products.

Relevance to autonomous legal systems

An autonomous legal ecosystem could similarly leverage dominance from one market into another.

For example:

Dominant legal database

↓

proprietary AI assistant

↓

document-management system

↓

legal workflow ecosystem

If interoperability with competing AI systems is restricted, the incumbent could use its position in the first market to strengthen its position downstream.

Case 6 — Google Shopping

Principle

The Google Shopping litigation concerned Google's treatment of its own comparison-shopping service within its general search results.

The case is important for the broader concept of self-preferencing by a dominant platform.

Relevance to autonomous legal systems

Imagine an AI legal platform answering:

"Which legal research database should I use?"

If the AI systematically recommends its affiliated database or legal-service provider while suppressing rivals, competition authorities could investigate:

• self-preferencing;
• discriminatory ranking;
• leveraging;
• foreclosure;
• transparency of ranking criteria.

The critical issue is not simply that an AI makes recommendations, but whether dominant market power is being used to disadvantage competing products.

20. Additional Relevant Case: United States v. Google

Modern digital-platform jurisprudence concerning Google's search and advertising businesses is also relevant to autonomous legal systems.

The underlying principles include:

• exclusionary agreements;
• distribution advantages;
• default positioning;
• network effects;
• barriers to entry;
• leveraging of platform power.

These concepts can be adapted to AI legal ecosystems where an incumbent controls:

• default AI assistants;
• legal databases;
• operating systems;
• enterprise software;
• cloud infrastructure.

21. Indian Competition-Law Perspective

India's Competition Act, 2002 prohibits anti-competitive agreements under Section 3 and abuse of dominant position under Section 4, while combination provisions address mergers and acquisitions capable of causing an appreciable adverse effect on competition.

An important Indian authority is:

Thupili Raveendra Babu v. Bar Council of India, Case No. 50/2020

The CCI examined allegations concerning the Bar Council of India's role in legal education and whether it could qualify as an enterprise for purposes of the Competition Act.

The case demonstrates an important preliminary question for competition law involving legal institutions:

Is the relevant body carrying on an economic activity falling within competition-law jurisdiction?

The CCI specifically considered the definition of "enterprise" under Section 2(h).

This becomes significant for autonomous legal systems because an AI-enabled legal ecosystem may combine:

• commercial activity;
• professional regulation;
• public legal information;
• statutory functions.

The correct competition-law analysis therefore requires separation of economic/commercial activity from sovereign or regulatory functions.

22. Autonomous Legal Systems and Section 3 of the Indian Competition Act

Potential Section 3 concerns include:

Horizontal agreements

Competing legal-AI providers could potentially coordinate:

• prices;
• access fees;
• licensing terms;
• API restrictions.

Vertical agreements

Potential concerns include:

• exclusive supply;
• exclusive distribution;
• tying;
• bundling;
• refusal to deal;
• restrictions on interoperability.

Algorithmic coordination

If competing legal-service providers employ autonomous pricing agents that independently coordinate market behaviour, the evidence would need to establish the legally relevant form of agreement or concerted conduct rather than merely observing parallel prices.

23. Autonomous Legal Systems and Section 4

Section 4 becomes particularly important when one company controls a critical legal-AI ecosystem.

Potential forms of abuse include:

A. Denial of market access

Refusing competitors access to critical legal information.

B. Discriminatory access

Giving the incumbent's AI system better access to data or APIs.

C. Predatory pricing

Using subsidised AI services to eliminate competitors.

D. Tying

Making legal research access conditional on purchasing an AI product.

E. Leveraging

Using dominance in legal databases to establish dominance in AI legal services.

F. Self-preferencing

Giving affiliated legal services preferential AI recommendations.

24. Competition Concerns in Autonomous Legal Agents

The next generation of autonomous legal systems may not simply answer questions.

They may:

• negotiate contracts;
• communicate with counterparties;
• select lawyers;
• file documents;
• monitor regulatory changes;
• initiate compliance actions;
• select arbitration mechanisms;
• recommend litigation strategies.

This creates a new competition concern:

The AI itself can become the decision-making gateway through which consumers access legal markets.

If a dominant AI agent determines which lawyers, legal databases, arbitrators or legal products users see, the system can exercise considerable gatekeeper power.

25. The "AI Legal Gatekeeper" Problem

A useful analytical model is:

Legal Data

↓

AI Model

↓

Recommendation Engine

↓

Legal Service Provider

↓

Consumer

The greatest competition concern may arise at the recommendation layer.

An autonomous system could determine:

• which lawyer is recommended;
• which precedent is displayed;
• which contract template is used;
• which legal product is purchased;
• which dispute-resolution mechanism is selected.

Consequently, competition law may need to scrutinise algorithmic intermediation, not merely traditional market shares.

26. Data Portability

Users should potentially be able to transfer:

• research histories;
• annotations;
• prompts;
• contract libraries;
• AI workflows;
• internal knowledge bases.

Restrictions on portability can produce substantial switching costs.

From a competition perspective, interoperability and portability can therefore become important non-price competitive parameters.

27. Transparency and Auditability

Competition authorities investigating autonomous legal systems may need access to:

• ranking logic;
• API-access policies;
• training-data sources;
• pricing algorithms;
• recommendation outputs;
• internal communications;
• model-change histories.

Traditional evidence-gathering methods may be insufficient because anti-competitive behaviour can emerge from complex AI systems rather than explicit corporate instructions.

28. Remedies

Potential competition remedies include:

1. Data-access remedies

Require reasonable access to certain datasets.

2. API interoperability

Require technical compatibility with rival systems.

3. Non-discrimination

Prevent preferential treatment of affiliated AI services.

4. Data portability

Permit customers to transfer their legal information and workflows.

5. Structural separation

In exceptional circumstances, separate:

• legal-data infrastructure;
• AI services;
• downstream legal-service platforms.

6. Merger remedies

Require divestiture or licensing of critical assets.

7. Algorithmic monitoring

Require independent audits of ranking and recommendation systems.

8. Compliance monitoring

Require continuing supervision where ordinary behavioural commitments may be difficult to verify.

29. Efficiency Defences

Autonomous legal systems can generate substantial pro-competitive efficiencies:

• lower legal research costs;
• faster dispute resolution;
• increased access to justice;
• lower compliance costs;
• better legal-document accuracy;
• greater competition from small law firms;
• increased innovation;
• improved legal-service availability.

Therefore, competition analysis should not assume that technological concentration is automatically harmful.

The central question is whether the conduct protects legitimate efficiencies or unnecessarily excludes competitors.

30. Key Competition-Law Test

A useful analytical framework is:

Step 1 — Define the market

Identify:

• AI legal research;
• legal databases;
• AI drafting;
• legal workflow platforms;
• legal-data infrastructure.

Step 2 — Identify market power

Examine:

• market share;
• data advantages;
• network effects;
• switching costs;
• entry barriers;
• technological superiority.

Step 3 — Identify conduct

Determine whether the conduct involves:

• tying;
• bundling;
• exclusivity;
• refusal to supply;
• self-preferencing;
• discrimination;
• predatory pricing;
• algorithmic coordination.

Step 4 — Establish foreclosure

Ask:

Does the conduct materially restrict competitors' ability to compete?

Step 5 — Examine efficiencies

Consider:

• innovation;
• security;
• quality;
• accuracy;
• privacy;
• cost savings.

Step 6 — Consider remedies

Possible remedies include:

• access;
• interoperability;
• non-discrimination;
• portability;
• licensing;
• behavioural commitments;
• structural remedies.

31. Comparative Case-Law Table

32. Major Emerging Issues

The next generation of competition litigation involving autonomous legal systems is likely to concern:

1. AI legal-data monopolies;
2. exclusive licensing of case-law datasets;
3. AI training-data foreclosure;
4. legal-database/API access;
5. AI self-preferencing;
6. algorithmic ranking discrimination;
7. AI-assisted cartelisation;
8. bundling of legal databases and AI assistants;
9. acquisition of nascent legal-AI competitors;
10. interoperability obligations;
11. legal-AI ecosystem lock-in;
12. algorithmic transparency;
13. competition between AI-generated and human legal services; and
14. competition effects of autonomous legal agents acting as market intermediaries.

33. Conclusion

Autonomous Legal Systems represent a significant evolution of legal technology because they combine legal information, artificial intelligence, automation and decision-making within a single ecosystem.

Their competition implications extend beyond ordinary software markets. The critical competitive assets may include legal data, case-law databases, citation networks, proprietary taxonomies, AI models, APIs, user data and distribution channels.

The principal competition-law risks are therefore:

data concentration + AI concentration + interoperability control + network effects + algorithmic recommendations + ecosystem lock-in.

The most important doctrinal tools are likely to include abuse of dominance, refusal to supply, essential-facilities principles, tying, self-preferencing, exclusive dealing, interoperability, algorithmic collusion and merger control.

The cases of United States v. Thomson Corporation, Thomson Reuters v. ROSS Intelligence, IMS Health, Bronner, Microsoft and Google Shopping collectively provide useful doctrinal foundations for analysing these emerging problems. The Indian framework under Sections 3 and 4 of the Competition Act can accommodate many of these concerns, although autonomous legal systems will require competition authorities to adapt traditional market-definition, evidence and remedy techniques to AI-driven ecosystems.