Competition Law and AI-Controlled Attention Allocation Markets

1. Introduction

“AI-controlled attention allocation markets” describes digital markets in which artificial intelligence determines what users see, in what order, how often, and for how long. Examples include search rankings, social-media feeds, recommendation systems, app-store rankings, advertising systems, video recommendations, news feeds, and AI assistants that select or prioritize information.

From a competition-law perspective, attention can function as an economically valuable and scarce resource. Businesses compete not only for money but also for visibility, clicks, engagement, user time, and access to consumers. When a dominant platform controls the AI system allocating that attention, it may effectively control an important route between businesses and customers.

There is not yet a large body of decided cases specifically labelled “AI-controlled attention allocation.” Therefore, the most useful precedents are cases concerning ranking, defaults, self-preferencing, platform access, tying, recommendation visibility, interoperability and control over digital distribution. These principles can be applied by analogy to AI-driven markets.

2. What Is an AI-Controlled Attention Allocation Market?

Suppose a platform has millions of pieces of content but a user can realistically examine only twenty. An AI system determines which twenty appear.

The algorithm may consider:

• previous clicks and searches;
• viewing time;
• purchases;
• location or contextual information;
• predicted interests;
• advertising bids;
• popularity;
• engagement probability;
• platform commercial objectives; and
• relationships between the platform and content providers.

Consequently, the platform is not merely hosting information. It can become an attention allocator.

Competition concerns become particularly significant where businesses cannot realistically reach consumers without passing through that allocation mechanism.

3. Competition-Law Importance of Attention

Traditional competition law frequently focuses on prices, output and market shares. AI attention markets require examination of additional competitive dimensions.

A service can have a monetary price of zero while competition occurs through:

User attention: platforms compete for the time users spend with their services.

Data: additional attention generates behavioural information that can improve AI systems.

Advertising: advertisers pay platforms for access to user attention.

Visibility: sellers and content providers compete for placement before consumers.

Quality and innovation: businesses may compete through recommendation quality, privacy, relevance and user experience.

This produces a potentially self-reinforcing mechanism:

More users → more attention → more interaction data → better AI → better targeting/recommendations → more users.

Competition law becomes relevant when firms use exclusionary conduct rather than legitimate competition on the merits to protect such an advantage.

4. Market Definition

Defining the relevant market can be particularly difficult.

Authorities might investigate separate markets for:

• general search;
• social networking;
• short-form video;
• online display advertising;
• search advertising;
• app distribution;
• recommendation services;
• AI assistants;
• content-discovery services; or
• specialised advertising technology.

A broad “attention market” should therefore not automatically be assumed.

The legal question remains whether consumers or advertisers regard different services as sufficiently substitutable.

For example, watching entertainment videos may consume the same person's time as searching the web, but that does not necessarily mean video platforms and general search engines belong to one antitrust market.

5. Main Competition Problems

A. Algorithmic Self-Preferencing

A vertically integrated platform can operate the marketplace while simultaneously competing against businesses dependent upon it.

Its AI system could rank:

1. the platform's own product;
2. affiliated products;
3. preferred commercial partners; and
4. independent competitors.

Self-preferencing is not automatically unlawful under every competition regime. The legal issue generally depends on dominance, the particular conduct, market circumstances, competitive effects and the governing legal test.

However, systematic manipulation of an important gateway can become an exclusionary concern.

B. AI Ranking Discrimination

AI systems may give different businesses substantially different exposure.

Suppose two sellers offer similar products but an algorithm consistently gives the platform's affiliated seller significantly greater visibility.

Competition authorities could investigate whether ranking criteria are genuinely connected with relevance, quality or efficiency, or instead operate as an exclusionary mechanism.

The difficult question is separating legitimate algorithmic optimization from anticompetitive discrimination.

C. Defaults and Choice Architecture

Default settings are especially important because many consumers do not change them.

An AI assistant could automatically choose:

• one search provider;
• one shopping service;
• one mapping provider;
• one music platform; or
• one booking provider.

A dominant firm might therefore obtain substantial traffic without consumers actively selecting it.

The U.S. Google search litigation is especially important here. A federal court found Google liable under Section 2 of the Sherman Act for maintaining monopolies in general search services and general search text advertising through exclusionary distribution agreements. In 2025, remedies included restrictions on certain exclusive distribution arrangements and requirements concerning access to specified search data and syndication services.

D. Attention Foreclosure

A dominant attention allocator could technically allow competitors onto its platform while making them practically invisible.

For example:

• competitor content could appear much lower in rankings;
• notifications could favour the platform's services;
• recommendation systems could rarely recommend rivals;
• competing applications could receive inferior placement; or
• AI-generated answers could replace links that previously generated traffic for independent providers.

This creates the concept of attention foreclosure: rivals remain legally present but lose meaningful access to users.

E. Tying and Bundling

A dominant platform could combine its AI attention mechanism with another service.

For example:

access to a major social platform + automatic exposure to its marketplace.

The European Commission's Meta/Facebook Marketplace decision illustrates the traditional competition principle. The Commission concluded that tying Facebook Marketplace to Facebook's personal social network constituted an abuse of Meta's dominant position under Article 102 TFEU and Article 54 EEA.

This principle could become important where AI recommendations automatically direct users toward another service owned by the platform.

F. Data Advantages and Feedback Loops

AI quality often improves with access to substantial interaction data.

A powerful platform may receive:

Attention → behavioural data → improved predictions → increased engagement → additional attention.

This does not make scale or superior AI unlawful.

The competition issue becomes stronger where exclusionary agreements, discriminatory access rules or other potentially unlawful practices prevent competitors from developing sufficient scale.

Data can therefore function as both an input into AI and a consequence of controlling attention.

G. Advertising Competition

Attention has direct commercial value in advertising-supported markets.

AI determines:

• which advertisement appears;
• where it appears;
• which consumer receives it;
• predicted conversion probability;
• advertising price; and
• frequency.

If one company controls the consumer interface, important data and advertising infrastructure, authorities may examine whether that integration permits exclusionary discrimination or unfair conditions.

The Commission's Meta Marketplace decision, for example, also addressed conditions imposed on advertising customers in the market for online display advertising on social-media platforms.

H. Acquisitions of Attention Competitors

Established platforms may acquire rapidly growing services before they become significant competitive constraints.

The competition concern is sometimes described as acquiring an emerging or nascent competitor.

In the FTC's Meta litigation, the agency alleged that Facebook maintained monopoly power partly through acquisitions of Instagram and WhatsApp and restrictions affecting developers. However, it is important to distinguish allegation from established liability: the district court ruled for Meta in November 2025, and the FTC filed an appeal in January 2026.

For future AI markets, merger authorities may similarly examine acquisitions of emerging recommendation engines, AI assistants or other services capable of redirecting substantial user attention.

6. Important Case Laws

Case 1: Google Shopping — Google and Alphabet v European Commission

This is one of the strongest precedents for understanding competition involving algorithmically allocated visibility.

The Commission found that Google had favoured its own comparison-shopping service in general search results while rival comparison-shopping services were disadvantaged.

The litigation ultimately reached the Court of Justice of the European Union, which in 2024 upheld the Commission's decision.

Importance for AI attention markets

The case demonstrates that control over ranking and visibility can have competition-law significance.

An AI system does not need literally to prevent competitors from operating. Reduced visibility can potentially affect competitive opportunities where access to the platform's user base is commercially important.

The principle can therefore inform analysis of AI recommendation engines that favour affiliated services.

Case 2: United States v Google LLC — Search Distribution

This U.S. litigation concerned Google's agreements for distribution of its search engine.

The U.S. District Court concluded in August 2024 that Google violated Section 2 of the Sherman Act by maintaining monopolies in general search services and general search text advertising through exclusionary distribution agreements.

The case was particularly concerned with default distribution arrangements. The Justice Department described arrangements through which Google obtained preset/default status across important search access points.

In September 2025, the court imposed remedies including restrictions on certain exclusive arrangements and requirements for specified search index/user-interaction data and search syndication access.

Importance

The case illustrates that control over defaults can influence attention allocation.

Future AI assistants could create similar issues if one dominant ecosystem automatically routes user questions, shopping requests or other activities toward preferred providers.

Case 3: European Commission v Google — Android

The Commission's Android proceedings concerned contractual restrictions associated with Android devices.

Among other findings, the Commission determined that licensing arrangements made access to the Play Store conditional on pre-installation of Google Search and, in relevant arrangements, default placement. It also addressed pre-installation of Chrome.

Importance

The case illustrates the competition significance of:

• defaults;
• pre-installation;
• prominent placement;
• tying;
• distribution control.

These principles translate naturally to AI-controlled interfaces.

For example, an operating system's AI assistant could potentially become the new default gateway through which consumers discover online services.

Case 4: Microsoft Corp. v United States

The historic Microsoft antitrust litigation concerned Microsoft's conduct surrounding Windows and Internet Explorer.

The case established important principles concerning monopoly maintenance, technological tying and restrictions that make it harder for competing technologies to obtain effective distribution.

Importance

Microsoft is particularly relevant because the central issue was not simply product price.

The dispute involved control over an important technological gateway.

AI assistants, operating systems and recommendation systems can similarly operate as gateways. A company controlling that gateway could potentially disadvantage technologies that threaten its existing market position.

The U.S. Justice Department itself expressly referenced the Microsoft precedent when bringing its Google search case, noting its relevance to preinstallation, defaults and restrictions on distribution.

Case 5: European Commission v Meta — Facebook Marketplace

In November 2024, the European Commission found that Meta had infringed Article 102 TFEU by tying Facebook Marketplace to Facebook's personal social network and through certain conditions imposed on advertising customers.

Importance

This provides an important analogy for AI attention markets.

A platform with an enormous existing audience can potentially give a related service immediate exposure that independent competitors cannot easily reproduce.

With AI, this effect could become even more powerful because recommendation engines can continuously decide when and where an affiliated service appears.

Case 6: European Commission v Apple — Music Streaming / Anti-Steering

The European Commission's Apple music-streaming proceedings concerned App Store rules restricting music-streaming developers from informing iOS users about alternative subscription options.

The Commission concluded that Apple's anti-steering provisions were not necessary for the objectives Apple relied upon and did not strike an appropriate balance with the interests of music-streaming providers and iOS users.

Importance

Steering determines where consumer attention goes after entering a platform.

AI interfaces could create a more sophisticated version of the same issue. An AI assistant might answer a user's request without revealing competing purchasing channels or might systematically route transactions toward the platform's preferred option.

The broader issue is whether platform rules restrict competitors' ability to communicate effectively with customers.

Case 7: FTC v Facebook/Meta

The FTC alleged that Facebook unlawfully maintained monopoly power in personal social networking through conduct including acquisitions of Instagram and WhatsApp and restrictions affecting software developers.

The amended complaint also alleged that developer-access restrictions hindered applications that might develop into competitive threats.

The procedural status matters. The district court ruled for Meta in November 2025, and as of the FTC's January 2026 announcement the agency had appealed that judgment.

Importance

The case illustrates how competition authorities may examine:

• network effects;
• acquisitions of emerging competitors;
• interoperability restrictions;
• access to platform infrastructure; and
• competition for user engagement.

These considerations are directly relevant where AI platforms compete to become the primary interface through which users allocate their online attention.

Case 8: Apple and Meta under the EU Digital Markets Act

Although the DMA is a regulatory regime rather than simply traditional Article 102 competition litigation, its enforcement provides additional insight into attention and user choice in digital markets.

In April 2025, the European Commission found Apple in breach of the DMA's anti-steering obligation and Meta in breach concerning consumer choice over a service using less personal data. The Commission imposed fines of €500 million and €200 million respectively.

Importance

AI attention systems raise closely related issues of effective choice.

Formal availability of alternatives may have limited competitive value where interface architecture or algorithmic design prevents consumers from realistically discovering or selecting them.

7. Legal Tests Applied to AI Attention Allocation

Competition authorities would generally need more than evidence that an algorithm produces unequal outcomes.

A structured investigation would examine:

First, market power. Does the platform possess dominance or monopoly power in a properly defined relevant market?

Second, conduct. Is the challenged ranking, default, contractual restriction, tying arrangement or access policy attributable to the undertaking?

Third, exclusion. Does the conduct materially restrict competitors' ability to compete?

Fourth, causation and effects. Is reduced visibility actually capable of harming the competitive process?

Fifth, objective justification or efficiency. Does the platform have legitimate reasons such as cybersecurity, relevance, fraud prevention, quality improvement or technical integration?

Sixth, proportionality where applicable. Could the legitimate objective reasonably be achieved through less restrictive measures?

8. Measuring Market Power in Attention Markets

Ordinary revenue market share may be insufficient.

Authorities could consider metrics such as:

• monthly active users;
• daily active users;
• average user time;
• search/query volume;
• recommendation impressions;
• advertising impressions;
• advertiser expenditure;
• click-through traffic;
• switching rates;
• default status;
• multi-homing;
• access to behavioural data; and
• dependency of businesses on platform-generated traffic.

This matters because an apparently “free” service may control extremely valuable commercial attention.

9. AI Transparency and Competition Evidence

Algorithmic opacity creates an evidential challenge.

A competitor may know that its traffic suddenly declined but not know whether this resulted from:

• legitimate relevance improvements;
• normal algorithm changes;
• consumer preferences;
• quality problems;
• spam enforcement;
• commercial arrangements; or
• intentional self-preferencing.

Competition investigations may therefore require examination of internal evidence such as ranking policies, experiments, model objectives, business rules and traffic effects.

However, competition law should not assume that every ranking change harming one competitor is anticompetitive. Algorithms necessarily differentiate among content.

10. Personalized Rankings Create a Special Problem

Traditional search rankings might produce roughly similar results for many users.

AI can produce a unique ranking for each person.

User A may see competitor X first.

User B may see the platform's own product first.

User C may never see competitor X.

Consequently, average rankings may conceal discriminatory effects.

Future competition analysis may increasingly examine distributional exposure—how visibility is allocated across millions of individualized recommendations rather than looking only at one public ranking page.

11. Generative AI and the “Zero-Click” Problem

Generative AI creates another possible competitive issue.

Traditional search:

Question → search results → external website.

AI interface:

Question → AI-generated answer → no external website visit.

The AI provider may therefore become both:

1. the gateway to information; and
2. the destination consuming the user's attention.

This does not automatically constitute an antitrust violation. But where a firm possesses substantial market power, authorities may investigate whether practices concerning access, attribution, ranking, distribution or affiliated services unlawfully foreclose competitors.

12. Potential Remedies

Where competition authorities establish an infringement, possible remedies—depending on jurisdiction and the violation—could include:

• ending exclusionary default arrangements;
• prohibiting discriminatory ranking practices;
• allowing effective steering to competing services;
• interoperability obligations;
• access to specified data;
• restrictions on tying;
• changes to contractual conditions;
• greater user choice over defaults;
• merger remedies; and
• behavioural monitoring.

The Google search remedies illustrate the direction such intervention can take: the 2025 U.S. judgment restricted specified exclusive distribution arrangements and required certain search-data and syndication access.

13. Key Competition-Law Principle

The central issue is not that AI algorithms allocate attention. Every search engine, recommendation system and social platform must rank or select information somehow.

The competition-law concern arises where a company with substantial market power allegedly uses control over that allocation mechanism to protect or extend market power through exclusionary conduct rather than competition on the merits.

Google Shopping demonstrates the importance of ranking and visibility; Google Search and Android demonstrate the power of defaults and distribution; Microsoft illustrates gateway foreclosure; Meta Marketplace demonstrates tying and leveraging; Apple illustrates restrictions on steering; and the Meta monopolization litigation shows how acquisitions, network effects and interoperability restrictions can become part of a broader monopolization theory.

AI therefore does not require competition law to abandon traditional concepts. Instead, it changes the technological mechanism through which familiar issues—foreclosure, tying, discrimination, defaults, network effects, vertical integration and control over distribution—may occur.