Competition Law and Competition Implications of Universal AI Assistants

1. Introduction

Universal AI assistants are AI systems designed to perform a wide range of activities through a single interface. Instead of using separate applications for search, shopping, travel, banking, communication, productivity, entertainment, education, coding and other services, a consumer may interact with one AI assistant that coordinates many of these functions.

Examples of functions potentially performed by a universal AI assistant include:

answering questions;

searching the internet;

recommending products;

booking travel;

ordering goods;

making payments;

managing calendars;

drafting communications;

writing software;

controlling smart devices;

accessing financial services;

selecting third-party applications;

negotiating transactions through AI agents.

From a competition-law perspective, universal AI assistants are important because they may become gateways between consumers and large numbers of businesses.

The central competition question is:

If one AI assistant becomes the primary gateway through which consumers discover, select and purchase products and services, can control over that gateway be used to create or protect market power in adjacent markets?

This issue connects AI competition with established doctrines concerning:

digital platforms;

monopolisation;

self-preferencing;

tying and bundling;

vertical foreclosure;

essential facilities;

data advantages;

network effects;

interoperability;

exclusionary contracts;

mergers and potential competition.

2. Meaning of a Universal AI Assistant

A universal AI assistant can be understood as:

A general-purpose AI interface capable of performing or coordinating multiple functions across different markets through natural-language or multimodal interaction.

A simplified structure is:

Consumer → Universal AI Assistant → Search / Apps / Commerce / Payments / Services

The assistant may therefore sit between the consumer and thousands of businesses.

This gives it potentially significant intermediation power.

3. Why Universal AI Assistants Are Different From Traditional Software

Traditional software generally performs a defined function.

For example:

search engine → search;

map application → navigation;

shopping application → shopping;

email application → communication.

A universal AI assistant can potentially combine all these functions.

Therefore:

One AI interface may compete simultaneously with multiple categories of applications.

This creates unusual competition-law problems because market boundaries become less clear.

4. Competition-Law Significance

Universal AI assistants can affect competition through at least ten mechanisms:

control over consumer access;

self-preferencing;

vertical integration;

tying and bundling;

data accumulation;

network effects;

switching costs;

exclusionary contracts;

acquisitions of emerging competitors;

algorithmic coordination.

5. Universal AI Assistants as Gatekeepers

A universal AI assistant may become a gatekeeper if consumers increasingly rely on it to access other businesses.

For example:

Consumer asks AI → AI recommends hotel → AI chooses airline → AI completes booking.

The consumer may never visit:

the hotel website;

the airline website;

the travel comparison website.

The AI assistant therefore controls an important stage of the competitive process.

This creates a possible shift:

Traditional model

Business → Search engine → Consumer

AI-mediated model

Business → AI assistant → Consumer

The AI assistant may therefore influence which businesses consumers see and select.

6. Market Power of the Interface

The interface itself can become a competitive asset.

A universal assistant may control:

rankings;

recommendations;

search results;

product selection;

app selection;

payment options;

transaction execution.

Consequently, the assistant may possess intermediation power even where it does not manufacture the products being sold.

7. Self-Preferencing

One of the most important competition concerns is self-preferencing.

Suppose an AI company operates:

an AI assistant;

a shopping service;

a travel service;

a payment service;

a cloud service.

The AI assistant might recommend its own services more frequently than rival services.

For example:

User: "Find me a hotel in Delhi."

The assistant could systematically favour a hotel-booking service owned by the same corporate group.

The issue is not simply that the firm owns both businesses.

The concern is whether control over the AI interface is used to disadvantage competing providers.

8. Google Shopping as a Relevant Case

In Google Search (Shopping), Case AT.39740, the European Commission found that Google had abused a dominant position by favouring its own comparison-shopping service in general search results.

The case is highly relevant to universal AI assistants because the fundamental issue concerns control over a digital gateway.

Competition principle

A dominant intermediary may face competition-law scrutiny when it uses control over an important access point to favour its own downstream service.

Application to AI

A universal AI assistant could potentially become the equivalent of a recommendation gateway.

If it systematically favours:

its own shopping service;

its own travel service;

its own financial product;

competition authorities may investigate whether the conduct excludes rivals.

9. Tying and Bundling

Universal AI assistants can combine numerous services into one package.

For example:

AI assistant + search + email + cloud + browser + shopping + payments

Bundling can produce legitimate efficiencies.

Consumers may benefit from:

convenience;

integration;

lower transaction costs;

personalization.

However, competition concerns may arise if a dominant firm conditions access to one service on acceptance of another service.

This raises traditional tying and bundling questions.

10. Microsoft Case

The leading authority is United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001).

Microsoft possessed substantial power in the PC operating-system market and integrated Internet Explorer into Windows.

The case examined whether Microsoft used its operating-system position to restrict competition in the browser market.

Relevance to universal AI assistants

The same conceptual issue can arise if an AI assistant is integrated into a dominant:

operating system;

browser;

mobile ecosystem;

cloud platform.

A company could potentially use its position in one market to make its AI assistant the default gateway to another market.

Principle

Control over a technologically important platform can potentially be leveraged into an adjacent market through integration, defaults or exclusionary conduct.

11. Google Android

The European Commission's Google Android decision provides another important example.

The Commission examined Google's practices involving Android, including contractual arrangements concerning search and browser distribution.

The case illustrates how:

defaults;

tying;

platform integration;

distribution restrictions

can affect competition in adjacent markets.

AI application

Suppose a universal AI assistant becomes integrated into a dominant mobile operating system.

The platform owner could potentially make its AI assistant:

the default assistant;

the default search interface;

the default shopping interface;

the default application-selection system.

The competition question would be whether such arrangements restrict competing assistants or downstream services.

12. Data as a Competitive Advantage

Universal AI assistants may have access to enormous amounts of information about users':

searches;

purchases;

preferences;

communications;

travel;

interests;

application usage;

interactions with businesses.

This creates a potentially powerful data feedback loop:

More users → more interactions → more data → better personalization → more users

Such feedback can reinforce market concentration.

13. Amazon Marketplace Case

The European Commission's investigation concerning Amazon Marketplace provides an important precedent concerning platform data.

The Commission examined Amazon's use of non-public seller data to compete with sellers using its marketplace.

Relevance to AI assistants

A universal AI assistant could potentially receive commercially sensitive information from:

merchants;

travel providers;

developers;

banks;

service providers.

If the assistant operator uses such information to develop competing products, competition concerns could arise.

For example:

AI assistant learns which products are selling through independent retailers and then launches a competing private-label service.

The competitive issue would concern the use of commercially sensitive information by a platform that simultaneously intermediates transactions.

14. Network Effects

Universal AI assistants may exhibit powerful network effects.

Consider:

More users → more data → better recommendations → more businesses join → more services → greater consumer value → more users

This creates a feedback loop.

A successful assistant could therefore become increasingly difficult for competitors to challenge.

15. Multi-Sided Market

A universal AI assistant may operate a multi-sided ecosystem involving:

consumers;

application developers;

merchants;

advertisers;

service providers;

payment providers;

content providers.

The assistant may therefore control relationships between several groups.

Competition authorities must examine effects across the ecosystem rather than viewing the assistant as an ordinary standalone product.

16. Vertical Foreclosure

Suppose a company controls:

Cloud → AI model → AI assistant → marketplace

It could potentially disadvantage competing businesses by:

restricting cloud access;

giving preferential computing resources to its own assistant;

favouring its own marketplace;

restricting API functionality;

controlling recommendations.

This is a classic vertical foreclosure concern.

17. Essential-Input Problem

Universal assistants may depend on certain critical inputs:

advanced computing;

foundation models;

data;

app ecosystems;

operating systems;

payment networks.

If a dominant company controls a critical input and competes downstream, it may have incentives to restrict rivals' access.

This raises traditional questions concerning:

refusal to deal;

discriminatory access;

interoperability;

essential facilities;

raising rivals' costs.

18. Qualcomm Case

The Qualcomm competition cases illustrate the importance of exclusivity and access to important technological inputs.

The European Commission examined Qualcomm's conduct involving baseband chipsets and exclusivity arrangements.

Relevance to AI

AI infrastructure may similarly involve scarce inputs such as:

specialised processors;

computing capacity;

cloud infrastructure.

If a dominant supplier enters downstream AI markets and simultaneously uses restrictive arrangements with customers, competition authorities may examine whether rivals are being foreclosed.

19. Interoperability

Interoperability is particularly important for universal AI assistants.

An AI assistant may need access to:

third-party applications;

payment systems;

calendars;

shopping platforms;

databases;

communication services.

If the dominant assistant refuses to interoperate with competing services, consumers may become locked into its ecosystem.

Therefore, competition remedies may potentially include:

API access;

interoperability;

data portability;

non-discrimination.

20. Switching Costs

Users may become heavily dependent on a universal AI assistant because it learns:

personal preferences;

communication patterns;

purchasing habits;

workflows;

contacts;

schedules.

The longer the consumer uses the assistant, the more valuable its personalization may become.

This can create:

Personalization → switching cost → user retention → more data → better personalization.

Such a feedback loop may strengthen market power.

21. Data Portability

Data portability can reduce switching costs.

A consumer might be able to transfer:

preferences;

history;

contacts;

settings;

transaction records;

personal AI configurations

to another assistant.

This could make competition more contestable.

However, portability alone may not solve competition problems if the dominant firm also controls:

computing;

distribution;

applications;

operating systems.

22. Default Position

Defaults can be particularly powerful.

Suppose a smartphone automatically uses:

Assistant A

instead of:

Assistant B.

Many consumers may never actively change the default.

The AI provider therefore receives an enormous distribution advantage.

This resembles the competition concerns examined in historical cases involving:

Microsoft Windows;

Google Android;

search defaults.

23. United States v. Google — Search Distribution

The U.S. litigation involving Google search provides an important modern example of competition concerns surrounding distribution agreements and default placement.

The case examined Google's agreements and practices concerning distribution of its search service and whether they contributed to maintaining monopoly power.

Universal AI implication

If a dominant platform makes its own AI assistant the default across:

smartphones;

browsers;

operating systems;

search interfaces;

competition authorities may examine whether those arrangements prevent rival assistants from gaining meaningful scale.

24. AI Recommendations and Ranking

Traditional search engines provide a list of results.

Universal AI assistants may provide a single answer or recommendation.

This creates an important difference.

Search engine

Consumer sees:

10 competing results

AI assistant

Consumer may receive:

1 recommended product

The AI therefore exercises greater selection power.

This may make ranking and recommendation algorithms particularly important to competition.

25. Algorithmic Discrimination

An AI assistant may recommend different products to different consumers based upon:

price;

preferences;

commission;

commercial agreements;

predicted behaviour.

If the dominant assistant systematically disadvantages competitors, competition concerns could arise.

Authorities may therefore examine:

ranking criteria;

commercial incentives;

recommendation rules;

discriminatory treatment.

26. AI and Advertising

Universal AI assistants may replace traditional advertising/search models.

Instead of:

"Here are 20 advertisements."

the assistant might say:

"I recommend Product X."

This could shift economic power from advertisers to the AI intermediary.

If the assistant controls both:

recommendation;

advertising;

it could potentially manipulate visibility of competing businesses.

27. Commission-Based Recommendations

AI assistants may eventually receive commissions from:

retailers;

hotels;

airlines;

financial providers;

restaurants;

software providers.

This creates an important conflict:

Consumer objective

Find the most suitable product.

Platform objective

Maximise commission revenue.

Competition concerns could arise if commercial incentives distort recommendations and systematically disadvantage rivals.

28. Universal AI Assistants and Financial Services

An AI assistant might eventually compare and recommend:

bank accounts;

loans;

insurance;

investment products.

If the assistant also owns a financial-services platform, it could favour its own products.

This creates potential concerns involving:

self-preferencing;

tying;

discrimination;

data advantages;

conflicts of interest.

29. Universal AI Assistants and E-Commerce

The same issue arises in retail.

An AI assistant could become the primary product-selection mechanism.

Instead of searching multiple websites, consumers might ask:

"Buy me a laptop under ₹70,000 suitable for programming."

The AI chooses:

brand;

retailer;

model;

price;

seller.

This creates significant selection power.

The AI assistant could therefore influence competition even without owning the underlying products.

30. Universal AI Assistants and Travel

A consumer could ask:

"Book the cheapest convenient flight and hotel."

The AI could decide:

airline;

hotel;

booking platform;

payment provider.

This potentially transforms the assistant into a digital travel intermediary.

If it favours affiliated businesses, competition concerns may arise.

31. Universal AI Assistants and App Stores

An AI assistant could replace conventional app discovery.

Instead of browsing an app store, a user could say:

"Find me an application that edits videos."

The AI chooses which application to install or use.

This may give the AI control over:

application visibility;

downloads;

recommendations;

access to users.

The traditional app-store gatekeeper problem could therefore migrate to the AI interface.

32. Universal AI Assistants and App Developers

Developers may become dependent upon the assistant.

A developer could face:

API fees;

ranking rules;

access restrictions;

commission structures;

data requirements.

If the assistant becomes indispensable, its contractual conditions may become commercially significant.

33. Exclusionary Contracts

A universal AI assistant may enter contracts requiring businesses to:

use its payment system;

use its cloud service;

avoid rival assistants;

provide exclusive access;

give preferential pricing;

refrain from direct distribution.

Such arrangements may raise competition concerns depending upon market power, duration, foreclosure effects and jurisdiction.

34. Exclusive Dealing

Suppose an AI assistant tells hotels:

"You can be recommended by our assistant only if you do not participate in rival AI recommendation systems."

If the assistant has substantial market power, such exclusivity could potentially foreclose competitors.

This resembles traditional exclusive-dealing analysis.

35. Tying Multiple AI Services

A company could potentially require users to accept:

AI assistant + cloud storage + browser + search + payment service

as a single package.

Bundling can create efficiency benefits.

But where a dominant firm uses one market's power to force adoption in another market, competition concerns may arise.

36. Killer Acquisitions in AI

Universal AI assistants may face competition from small startups developing:

specialised AI agents;

new architectures;

better search;

personalised assistants;

autonomous shopping agents;

privacy-focused AI.

A dominant company may have an incentive to acquire these firms before they become serious competitors.

Competition authorities may therefore examine:

potential competition;

innovation;

technology;

talent;

intellectual property.

37. Meta/Facebook Litigation

The FTC's litigation involving Facebook/Meta provides an important example of scrutiny over acquisitions and the preservation of competition in digital markets.

The case illustrates why competition authorities may examine not merely current market shares but also whether acquisitions remove potential or emerging competitive constraints.

For universal AI assistants, the acquisition of an emerging assistant or agent may therefore deserve attention even if the target's present revenues are small.

38. Competition Between AI Assistants

There may eventually be several categories:

general-purpose assistants;

enterprise assistants;

personal assistants;

open-source assistants;

specialised assistants;

autonomous agents.

Competition may depend upon:

accuracy;

price;

privacy;

speed;

integration;

personalization;

interoperability.

A market with many nominal AI assistants may nevertheless be concentrated if most depend on the same underlying infrastructure.

39. Infrastructure Bottlenecks

Consider:

10 AI assistants

but all ten depend on:

2 cloud providers

and:

3 major AI-chip suppliers.

Competition at the assistant level may therefore be constrained by upstream concentration.

This illustrates why authorities should examine the entire AI value chain.

40. Cloud Bundling

A cloud provider may offer:

Cloud + foundation model + universal AI assistant.

Bundling can provide significant efficiency.

But if customers are effectively required to purchase the complete package, competing cloud or AI providers may be disadvantaged.

This could raise tying and foreclosure questions.

41. Open-Source Assistants

Open-source assistants may increase competition by:

reducing licensing costs;

increasing transparency;

facilitating customization;

reducing vendor lock-in.

However, open-source assistants may still depend upon proprietary:

computing;

chips;

app stores;

cloud services.

Thus, open source does not automatically eliminate market concentration.

42. Universal AI Assistants and Privacy

Privacy is not traditionally identical to competition law.

However, privacy can become a quality dimension of competition.

Consumers may prefer:

local processing;

limited data collection;

no behavioural tracking.

If a merger eliminates a privacy-focused competitor, competition on privacy may decline.

Therefore, non-price competition may be relevant.

43. Quality Competition

Competition among AI assistants can occur through:

accuracy;

safety;

privacy;

reliability;

transparency;

response speed;

personalization.

A dominant company may therefore harm competition without raising monetary prices.

For example:

If consumers receive less privacy or poorer service after rivals are excluded, competitive harm may occur even when the assistant remains free.

44. Innovation Competition

Innovation is especially important because AI technology changes rapidly.

Today's leading AI assistant may face tomorrow's technological substitute.

Competition authorities therefore need to consider:

R&D pipelines;

patents;

emerging architectures;

startups;

open-source alternatives;

potential entrants.

45. Universal AI Assistants and Consumer Choice

There is a potential paradox.

The assistant makes consumers' lives easier by reducing the need to compare many options.

But:

Convenience can reduce competitive visibility.

If the assistant presents only one recommendation, consumers may not know:

what alternatives existed;

why one product was selected;

whether another provider offered better terms.

This can increase the assistant's market power.

46. Transparency and Competition

Competition authorities may therefore be interested in:

ranking criteria;

commercial relationships;

affiliate relationships;

recommendation incentives;

access conditions.

However, competition law should not automatically require complete disclosure of proprietary algorithms.

The relevant question is whether insufficient transparency facilitates anti-competitive exclusion.

47. Interoperability as a Remedy

Potential remedies could include:

API interoperability

Allow competing assistants to access services.

Data portability

Allow users to transfer their data.

Application portability

Allow applications to function across assistants.

Non-discrimination

Require equal treatment of competing providers.

Choice screens

Allow users to select among competing assistants.

48. Structural Remedies

In serious cases, authorities may consider structural remedies such as:

divestiture;

separation of businesses;

restrictions on cross-ownership.

Structural remedies are generally more intrusive and may be considered where behavioural remedies cannot adequately preserve competition.

49. Indian Competition-Law Perspective

In India, universal AI assistants can potentially be analysed under the Competition Act, 2002.

Section 3 — Anti-competitive agreements

Potential issues include:

algorithmic coordination;

exclusive arrangements;

information sharing;

restrictive technology agreements.

Section 4 — Abuse of dominant position

Potential concerns include:

discriminatory access;

denial of market access;

tying;

leveraging;

predatory pricing;

unfair conditions.

Sections 5 and 6 — Combinations

Potential transactions include:

AI-company acquisitions;

cloud-AI mergers;

acquisition of AI startups;

vertical integration;

acquisition of datasets or technology.

50. Relevant Indian Competition Concepts

The CCI could potentially consider:

market share;

entry barriers;

network effects;

technological advantages;

data advantages;

consumer dependence;

switching costs;

vertical integration;

innovation;

countervailing buyer power.

Importantly, possession of a sophisticated AI assistant does not automatically establish dominance. The relevant market and statutory factors must be assessed.

51. Six Major Case-Law Principles

52. Key Competition Risks

The major risks can be summarized as:

A. Gateway power

One assistant becomes the primary route to consumers.

B. Self-preferencing

The assistant favours affiliated businesses.

C. Tying

AI services are tied to other products.

D. Data accumulation

The assistant obtains enormous behavioural datasets.

E. Network effects

More users make the assistant increasingly difficult to challenge.

F. Switching costs

Users become dependent on personalized AI ecosystems.

G. Vertical foreclosure

Infrastructure is used to disadvantage competing assistants.

H. Killer acquisitions

Emerging AI competitors are acquired before they scale.

I. Algorithmic coordination

AI systems may facilitate coordinated conduct.

J. Infrastructure concentration

Multiple assistants depend upon a small number of upstream providers.

53. Potential Pro-Competitive Effects

Universal AI assistants can also generate substantial competition benefits.

They may:

reduce search costs;

increase consumer choice;

make markets more accessible;

help small businesses reach customers;

reduce transaction costs;

improve price comparison;

enable personalized services;

facilitate entry;

increase productivity;

stimulate innovation.

Therefore, competition law should distinguish integration that creates genuine efficiencies from integration that unnecessarily excludes rivals.

54. The Central Competition Paradox

Universal AI assistants create a distinctive paradox:

The more convenient the assistant becomes, the more important its competitive neutrality may become.

If consumers independently compare dozens of websites, each business competes directly for attention.

If one AI assistant makes the decision for the consumer, businesses may instead compete for access to the AI's recommendation system.

The competitive battleground therefore moves:

From consumer attention → to AI intermediation.

55. Future Competition-Law Questions

Several questions are likely to become increasingly important:

1. Can an AI assistant be a gatekeeper?

Yes, potentially, if it becomes an important intermediary between users and businesses.

2. Can AI recommendations constitute self-preferencing?

Potentially, depending upon the assistant's market position and conduct.

3. Can AI assistants facilitate algorithmic collusion?

Potentially, depending upon how algorithms are designed, instructed and used.

4. Can an AI assistant's data become an entry barrier?

Yes, particularly where data creates substantial quality or personalization advantages that rivals cannot readily reproduce.

5. Can AI startups be important competitors despite low revenue?

Yes. Their future competitive significance may be greater than their current sales suggest.

56. Conclusion

Universal AI assistants have the potential to become a new layer of economic infrastructure. They may not merely provide information; they may determine which products consumers see, which services they select, which applications they use and which transactions they complete.

This creates several competition-law challenges.

The principal concern is the possibility of a transition from:

Many businesses competing for consumers

to:

Many businesses competing for access to one AI-controlled gateway.

Competition law therefore needs to pay particular attention to:

self-preferencing;

defaults;

tying and bundling;

data advantages;

interoperability;

switching costs;

vertical foreclosure;

exclusive arrangements;

AI infrastructure;

potential competition;

acquisitions of emerging AI firms;

algorithmic coordination.

The cases of Microsoft, Google Shopping, Google Android, Google Search, Amazon Marketplace, Qualcomm and Facebook/Meta provide established competition-law principles that can be applied to these emerging AI-market structures.

The fundamental principle is:

Universal AI assistants can increase consumer convenience and market efficiency, but where a single assistant becomes a critical gateway between consumers and competing businesses, competition law must examine whether control of that gateway is being used to preserve or extend market power.

Quick Revision Points

Universal AI assistants combine multiple services through one interface.

They may become important digital gateways.

Gateway control can create significant market power.

Self-preferencing is a major potential concern.

Defaults can reinforce AI-assistant dominance.

Bundling may produce efficiencies but can also create foreclosure.

Data can create powerful feedback loops.

Network effects can reinforce concentration.

Switching costs can lock consumers into one assistant.

AI assistants can influence recommendations and purchasing decisions.

Vertical integration can create foreclosure risks.

Cloud and computing concentration can affect AI competition.

Algorithmic coordination is an emerging competition issue.

AI acquisitions can eliminate potential competitors.

Microsoft illustrates platform leveraging.

Google Shopping illustrates self-preferencing.

Google Android illustrates tying, defaults and platform power.

Amazon Marketplace illustrates platform-data concerns.

Qualcomm illustrates exclusivity involving important technological inputs.

Facebook/Meta illustrates potential-competition concerns.

Indian analysis can involve Sections 3, 4, 5 and 6 of the Competition Act, 2002.

Interoperability and data portability can reduce lock-in.

Universal AI assistants can simultaneously create efficiency and concentration.

The key competition issue is whether AI intermediation remains open, contestable and non-discriminatory.