Global Risk-Pricing AI Systems and Economic Vulnerability Control

Introduction

Risk-pricing AI systems are algorithmic systems that determine, predict, or influence the price or availability of products and services according to perceived risk. They are increasingly used in insurance, lending, credit scoring, financial markets, employment, logistics, healthcare, cybersecurity, energy and digital platforms.

Examples include:

• AI credit-scoring and loan-pricing systems;
• insurance underwriting and premium-setting algorithms;
• dynamic pricing based on behavioural or location data;
• AI systems that price financial risk and allocate capital;
• algorithmic trading and liquidity-risk models;
• supply-chain and catastrophe-risk pricing;
• AI systems that determine deposits, collateral, guarantees or credit limits.

The competition-law problem becomes particularly important where risk pricing turns into economic vulnerability control—that is, where an AI system does not merely measure risk but determines who receives affordable access to credit, insurance, housing, employment, essential services or market participation.

The central legal question is:

When does legitimate automated risk assessment become a mechanism for exclusion, discrimination, coordinated pricing, exploitation or structural control of economically vulnerable groups?

1. Meaning of Economic Vulnerability Control

Economic vulnerability control occurs when an entity with substantial informational, technological or market power can use AI-generated risk assessments to determine the economic opportunities available to individuals or businesses.

The mechanism can be represented as:

Data → Risk Classification → AI Prediction → Risk Score → Price/Access Decision → Economic Consequence

For example:

Personal data → AI predicts default probability → borrower receives high-risk classification → higher interest rate → reduced borrowing capacity → reduced economic opportunity.

The concern is amplified when the AI system controls several interconnected decisions:

1. access to credit;
2. interest rate;
3. collateral requirements;
4. insurance premium;
5. credit limit;
6. payment terms;
7. employment-related economic opportunities;
8. access to digital marketplaces.

A single risk score can therefore become a gatekeeping mechanism for economic participation.

2. Global Competition-Law Dimensions

The issue intersects several areas of competition law.

A. Abuse of dominance

A dominant undertaking may use AI risk systems to:

• exclude competitors;
• discriminate against dependent customers;
• impose unfair pricing;
• deny access to essential data;
• leverage dominance from one market into another.

B. Algorithmic collusion

Several firms may independently use similar risk-pricing systems.

Even without an explicit agreement, algorithms may:

• observe competitors' prices;
• rapidly react to changes;
• converge on supra-competitive prices;
• reduce incentives to compete;
• facilitate tacit coordination.

C. Data advantages

Large platforms may possess extensive behavioural, transactional and financial datasets unavailable to rivals.

This can create a feedback loop:

More data → better risk prediction → better pricing → more customers → more data → stronger market position.

D. Exploitative pricing

AI may identify consumers with a high willingness or necessity to pay and charge them more.

The issue is especially serious where consumers have:

• low income;
• limited alternatives;
• poor credit histories;
• geographic constraints;
• high switching costs.

E. Discriminatory market access

AI models can indirectly reproduce discriminatory outcomes through apparently neutral variables such as:

• postcode;
• device characteristics;
• browsing behaviour;
• transaction history;
• employment patterns;
• social-network characteristics.

3. Risk Pricing as a Potential Essential-Input Problem

In digital markets, the critical input may not be money itself but risk information.

A dominant platform may control:

• transaction data;
• consumer behavioural information;
• payment histories;
• identity data;
• merchant data;
• credit-performance data.

If competitors cannot obtain comparable information, the dominant undertaking may possess an important competitive advantage.

The legal problem becomes:

Can competition law require access to data or risk infrastructure when exclusion prevents effective competition?

This potentially engages the principles associated with essential facilities, refusal to deal and discriminatory access.

4. Algorithmic Risk Pricing and Consumer Welfare

Traditional competition analysis often focuses on price.

AI risk systems complicate this approach because the competitive harm may occur through:

• higher risk premiums;
• reduced credit availability;
• exclusion from marketplaces;
• inferior service;
• reduced privacy;
• discriminatory treatment;
• reduced innovation;
• increased dependency.

Consequently, a transaction can remain nominally competitive while the AI system produces non-price harm.

For example:

Two lenders may compete aggressively for low-risk customers while simultaneously using automated systems to charge vulnerable borrowers substantially higher rates.

A conventional average-price analysis may understate the competitive harm.

5. Six Important Case Laws

1. United States v. Socony-Vacuum Oil Co. — U.S.

The Supreme Court treated concerted price-fixing as a fundamental antitrust violation.

Relevance

The case is important for AI risk pricing because algorithmic systems can potentially facilitate coordination among competitors.

Suppose several insurers independently deploy AI systems that continuously monitor competing premiums and automatically adjust their own prices.

If the systems facilitate an agreement or coordinated conduct, the underlying technology does not immunize the conduct from antitrust scrutiny.

Principle

Technology cannot convert unlawful price coordination into lawful competition.

Application

AI pricing systems therefore require scrutiny of:

• communications between competitors;
• common pricing software;
• shared algorithms;
• common data providers;
• algorithmic instructions;
• automated reactions to competitors.

6. United States v. Apple Inc. — E-books

The Apple e-books litigation concerned coordination affecting prices in the e-books market.

Relevance to AI

The case illustrates an important principle for technology-mediated pricing:

Competition authorities examine the economic arrangement and its effects rather than accepting technological or contractual structures at face value.

For AI risk pricing, companies cannot necessarily avoid antitrust scrutiny merely because pricing decisions are implemented through:

• software;
• automated instructions;
• machine-learning models;
• third-party platforms.

Broader lesson

Where AI becomes the mechanism through which competitors implement coordinated pricing, authorities can investigate the underlying human and organizational conduct.

7. FTC v. Amazon — U.S.

The Federal Trade Commission's litigation against Amazon concerns alleged conduct affecting competition in online retail.

Relevance

The case is particularly useful for understanding how a dominant digital platform can use technological infrastructure and pricing mechanisms to influence market outcomes.

AI-enabled pricing can become problematic when a platform simultaneously:

• operates the marketplace;
• collects extensive transaction data;
• controls seller access;
• determines ranking;
• influences prices;
• monitors competitor behaviour.

Competition concern

This creates a potential information-and-control feedback loop.

The platform may possess information unavailable to sellers and competing marketplaces.

Principle

Digital market power can involve more than conventional ownership of physical infrastructure.

Control over information, algorithms, visibility and access can itself generate competitive leverage.

8. Google Shopping — European Union

The European Commission's Google Shopping decision concerned Google's preferential treatment of its own comparison-shopping service.

Relevance to AI risk systems

The case demonstrates how an undertaking possessing substantial platform power can use an algorithmic system to influence downstream competitive conditions.

The concern was not simply that Google possessed an algorithm.

Rather, the issue was how the platform's algorithmic mechanisms affected competing services.

Application to risk pricing

An AI platform could similarly manipulate economic opportunities by:

• preferentially allocating favourable risk classifications;
• giving its own financial products better risk assessments;
• disadvantaging competing lenders;
• restricting competitors' access to relevant data.

Principle

Algorithmic neutrality is not guaranteed merely because decision-making is automated.

9. Slovak Telekom v European Commission — CJEU

The case concerned exclusionary conduct and access to telecommunications infrastructure.

Relevance

The case is significant for understanding when control over an important infrastructure or input can create competition-law obligations.

In AI risk markets, the relevant infrastructure may be:

• credit databases;
• financial-data platforms;
• insurance-risk databases;
• payment networks;
• identity systems;
• cloud-based AI infrastructure.

Application

If a dominant firm controls an indispensable risk-data infrastructure and uses that control to exclude rivals, competition authorities may examine the conduct under refusal-to-deal and exclusionary-abuse principles.

Principle

Control over strategically important infrastructure can translate into downstream market power.

10. Intel v European Commission — CJEU

The Intel litigation concerns exclusionary rebates and the assessment of whether conduct by a dominant undertaking can restrict competition.

Relevance to AI pricing

The broader lesson is that competition analysis should consider the actual competitive mechanism through which conduct can foreclose rivals.

For AI risk-pricing systems, authorities may need to examine:

• how the algorithm affects rivals;
• the proportion of customers affected;
• switching possibilities;
• alternative data sources;
• network effects;
• foreclosure duration;
• access to competing risk models.

Principle

The existence of a sophisticated pricing or incentive system does not eliminate the need for an effects-based assessment where appropriate.

11. Booking.com — European Competition Context

Competition authorities and courts have examined restrictions imposed by online booking platforms concerning hotel pricing.

Relevance to AI

Digital platforms increasingly combine:

• algorithmic ranking;
• demand prediction;
• dynamic pricing;
• personalised recommendations;
• commission optimisation.

A platform that controls both marketplace visibility and pricing intelligence can exert substantial influence over dependent businesses.

AI risk

If the same infrastructure determines:

risk → ranking → price → visibility → customer allocation

the platform may acquire a powerful form of vertical control.

Principle

Competition analysis increasingly considers the interaction between platform governance and commercial pricing.

12. Common Legal Issues Across These Cases

13. The Data–Risk–Power Feedback Loop

One of the most important emerging issues is the risk-data feedback loop.

Stage 1 — Data accumulation

A platform collects:

• transactions;
• searches;
• payments;
• purchases;
• location;
• device information;
• financial behaviour.

Stage 2 — Risk modelling

AI converts the data into predictions.

Stage 3 — Pricing

The platform modifies:

• interest;
• insurance premium;
• credit limit;
• deposit;
• access conditions.

Stage 4 — Behavioural response

Customers modify their behaviour because of the pricing.

Stage 5 — New data

The modified behaviour produces additional data.

Stage 6 — Model reinforcement

The AI model becomes increasingly powerful.

This creates:

Data → Prediction → Pricing → Behaviour → More Data → Greater Market Power

This is potentially a significant competition-law concern because competitors may be unable to reproduce the same learning environment.

14. Economic Vulnerability and Algorithmic Discrimination

AI risk pricing can produce proxy discrimination.

An algorithm may never explicitly use a protected characteristic but may use variables strongly correlated with it.

For example:

Variable → postcode → historical financial patterns → predicted risk → higher price

The company may argue that the model is purely statistical.

However, competition and regulatory analysis may examine the economic effect of the model.

The legal issue becomes particularly serious when vulnerable consumers cannot realistically switch providers.

15. Dynamic Risk Pricing and Exploitative Conduct

Dynamic pricing traditionally means that prices change according to demand.

AI can take this considerably further.

Instead of:

"What is the market demand?"

the system may ask:

"What is this particular consumer likely to accept?"

That can create individualised vulnerability pricing.

For example:

• financially distressed consumer → higher interest;
• emergency purchaser → higher price;
• consumer with few alternatives → higher premium;
• dependent business → worse contractual terms.

The competition-law concern is strongest where the firm possesses substantial market power.

16. AI Risk Models and Financial Stability

The issue is not limited to individual consumers.

If many financial institutions rely upon similar AI models, systemic vulnerability may arise.

For example:

Common training data → common risk model → common risk classification → simultaneous lending withdrawal

A shock could therefore cause multiple institutions to make similar decisions simultaneously.

This can create:

• credit contraction;
• asset-price declines;
• liquidity shortages;
• procyclical lending;
• market instability.

Thus, AI risk pricing can transform private risk assessment into systemic economic risk.

17. Competition Between Risk Models

Competition authorities may need to consider whether there is genuine competition between:

1. different AI models;
2. different data suppliers;
3. different risk infrastructures;
4. different financial platforms.

A market can appear competitive at the customer-facing level while the underlying risk infrastructure is highly concentrated.

For example:

100 lenders

may compete for customers while relying upon:

2 dominant risk-data providers

The apparent competition therefore rests upon a concentrated upstream infrastructure.

18. Possible Competition-Law Remedies

Authorities may consider several remedies.

Structural remedies

• divestiture;
• separation of platform and financial services;
• separation of data infrastructure;
• interoperability obligations.

Behavioural remedies

• non-discrimination;
• fair access to risk data;
• algorithmic auditing;
• transparency requirements;
• restrictions on self-preferencing;
• prohibition of discriminatory pricing.

Data remedies

• data portability;
• interoperable APIs;
• access to certain datasets;
• independent data trustees;
• privacy-preserving data sharing.

Algorithmic remedies

• independent auditing;
• model testing;
• explainability;
• monitoring for discriminatory outcomes;
• documentation of material model changes.

19. Relationship With AI Regulation

Competition law should not operate alone.

Risk-pricing AI can simultaneously raise questions under:

• competition law;
• financial regulation;
• consumer protection;
• data protection;
• equality/discrimination law;
• AI regulation;
• prudential regulation;
• cybersecurity law.

This produces a multi-regulator problem.

For example:

A credit-pricing algorithm could be simultaneously discriminatory, privacy-invasive, exclusionary and systemically risky.

A purely competition-law analysis might therefore miss important aspects of the harm.

20. Global Regulatory Approach

Different jurisdictions may approach the issue differently.

United States

Emphasis traditionally includes:

• Sherman Act;
• Clayton Act;
• FTC Act;
• sectoral financial regulation;
• consumer-protection enforcement.

European Union

Potentially relevant frameworks include:

• Articles 101 and 102 TFEU;
• Digital Markets Act;
• GDPR;
• AI regulation;
• financial-services regulation.

United Kingdom

Relevant institutions and frameworks include:

• Competition and Markets Authority;
• Financial Conduct Authority;
• UK competition law;
• UK data-protection rules;
• emerging AI regulatory arrangements.

India

Potential issues arise under:

• Competition Act, 2002;
• Digital Personal Data Protection framework;
• RBI financial regulation;
• consumer-protection law;
• sector-specific digital regulation.

21. Key Doctrinal Test

A useful legal framework is:

Step 1 — Identify the market

Is the relevant market:

• lending?
• insurance?
• credit scoring?
• risk-data provision?
• AI infrastructure?
• financial intermediation?

Step 2 — Identify the AI controller

Who controls:

• data;
• model;
• infrastructure;
• pricing;
• customer interface?

Step 3 — Determine market power

Examine:

• market share;
• data advantages;
• network effects;
• switching costs;
• barriers to entry;
• interoperability.

Step 4 — Identify the vulnerable group

Who is economically dependent?

• consumers;
• SMEs;
• merchants;
• borrowers;
• insurers;
• competing platforms.

Step 5 — Analyse the algorithm

Determine whether it:

• prices;
• ranks;
• excludes;
• coordinates;
• discriminates;
• self-preferences.

Step 6 — Establish competitive harm

Possible harm includes:

• foreclosure;
• excessive pricing;
• reduced choice;
• exclusion;
• innovation suppression;
• coordinated pricing.

Step 7 — Assess justification

The undertaking may invoke:

• legitimate risk management;
• fraud prevention;
• efficiency;
• financial stability;
• consumer protection.

The authority must distinguish genuine risk management from strategic exploitation of vulnerability.

22. Emerging Legal Concept: Algorithmic Economic Dependency

A particularly important future concept is algorithmic economic dependency.

A business or consumer may technically have a choice of providers but practically be unable to escape a dominant AI infrastructure.

For example:

Dominant risk database → dominant credit score → dominant lender ecosystem → restricted economic opportunity

The resulting power is different from traditional monopoly power because it operates through prediction and classification.

The dominant undertaking does not merely sell a product.

It determines the economic category into which the customer is placed.

23. Critical Competition-Law Question

The most difficult legal question is whether competition law should intervene only when AI risk pricing harms competition between undertakings, or whether it should also address the broader economic vulnerability created for consumers.

A modern approach increasingly recognises that competition may be undermined where dominant digital infrastructure allows firms to control:

• access;
• information;
• opportunity;
• price;
• visibility;
• risk classification.

Thus, economic vulnerability can become a competition issue when it is produced or amplified by market power.

Conclusion

Global risk-pricing AI systems represent a new form of economic power. Their importance goes beyond automated pricing because they can determine who is considered risky, who receives favourable economic terms, who obtains access to markets and who becomes economically dependent.

The six-plus authorities discussed above demonstrate several foundational principles:

1. AI cannot immunize coordinated pricing from antitrust law.
2. Algorithmic systems can constitute mechanisms of exclusion.
3. Control over critical data and infrastructure can create downstream market power.
4. Dominant digital platforms may influence competitive conditions through algorithms.
5. Risk assessment can become discriminatory or exploitative when combined with substantial market power.
6. Competition authorities increasingly need to examine data, algorithms and infrastructure together rather than treating pricing as an isolated decision.

The emerging legal principle can therefore be expressed as:

Where a powerful undertaking controls the data, algorithms and infrastructure used to price economic risk, competition law must examine not merely whether prices are high or low, but whether algorithmic risk classification itself has become a mechanism for exclusion, coordination, exploitation or economic dependency.