Civil Law And Algorithmic Wage Setting Fairness Litigation In Europe

1. Introduction

Algorithmic wage setting means using software, AI, automated scoring, data analytics, or algorithmic management to determine or influence how much a worker earns.

Examples include algorithms that determine:

hourly pay;

piece-rate payments;

bonuses;

commissions;

performance pay;

surge or dynamic remuneration;

delivery-worker compensation;

incentives;

deductions;

pay progression;

promotion-linked remuneration;

allocation of higher-paying assignments.

The legal problem arises when an algorithm produces wage differences that are:

discriminatory;

unexplained;

based on inaccurate data;

based on biased performance measures;

inconsistent with equal-pay rules;

contrary to contractual obligations;

insufficiently transparent;

indirectly discriminatory.

European law is particularly significant because Article 157 TFEU guarantees equal pay for men and women for equal work or work of equal value, while newer EU legislation increasingly requires transparency concerning how remuneration is determined. (EUR-Lex)

A crucial qualification is necessary: there is not yet a large body of CJEU judgments in which a court has directly ruled that an AI wage-setting algorithm itself unlawfully discriminated against workers. The strongest legal analysis therefore combines established equal-pay jurisprudence with the newer EU rules governing algorithmic management and pay transparency.

2. Meaning of Algorithmic Wage-Setting Fairness

The basic process is:

WORKER DATA → ALGORITHM → PERFORMANCE/RISK SCORE → PAY CALCULATION → WAGE DIFFERENCE → LEGAL REVIEW

The relevant data might include:

hours worked;

productivity;

sales;

customer ratings;

acceptance rates;

location;

availability;

experience;

absence records;

historical wages;

task completion speed;

employer evaluations;

algorithmically predicted performance.

The legal question is not simply:

"Did the computer calculate the wage?"

The important question is:

Was the resulting wage determined according to lawful, objective, transparent and non-discriminatory criteria?

3. EU Legal Framework

Several legal regimes may overlap.

A. Article 157 TFEU

Article 157 establishes equal pay between men and women for:

equal work; and

work of equal value.

The CJEU has confirmed that this principle can be relied upon directly in disputes between private individuals. Tesco Stores, C-624/19 is particularly important. (curia)

B. Directive 2006/54/EC

The Equal Treatment Directive provides the principal EU framework for gender equality in employment and pay.

It addresses:

direct discrimination;

indirect discrimination;

equal pay;

burden of proof;

remedies.

C. Pay Transparency Directive 2023/970

Directive (EU) 2023/970 is especially important for algorithmic wage-setting.

It requires employers to make accessible the criteria used to determine:

workers' pay;

pay levels;

pay progression.

Those criteria must be objective and gender neutral. (EUR-Lex)

The Directive also gives workers rights to obtain information concerning:

their individual pay;

average pay levels;

sex-disaggregated pay information for comparable categories;

additional clarification where information is incomplete or inaccurate. (EUR-Lex)

This is extremely relevant to algorithmic remuneration because an employer should increasingly be able to explain the criteria by which its remuneration system operates.

4. Algorithmic Management and Platform Work

Directive (EU) 2024/2831 on platform work is particularly important.

It expressly addresses automated monitoring systems and automated decision-making systems.

The Directive recognises that algorithms can perform functions traditionally performed by managers, including:

allocating tasks;

determining schedules;

evaluating performance;

providing incentives;

determining earnings;

applying adverse treatment.

(EUR-Lex)

Article 9 requires transparency concerning automated monitoring and automated decision-making systems. Article 10 requires human oversight, including personnel with sufficient competence, training and authority to override automated decisions. (EUR-Lex)

Thus, in platform work:

algorithmic wage setting is increasingly treated as an employment-law governance issue, not merely a software issue.

5. What Makes an Algorithmically Determined Wage Unfair?

An algorithmic wage system may create legal problems through several mechanisms.

1. Direct discrimination

Example:

Female worker → lower algorithmic bonus than male worker for the same qualifying performance.

2. Indirect discrimination

Example:

Algorithm heavily rewards uninterrupted availability, disproportionately disadvantaging workers with caring responsibilities.

3. Proxy discrimination

Example:

Location or historical customer ratings indirectly reproduce gender or ethnic disparities.

4. Historical-data bias

Example:

Historical wages were lower for women → algorithm learns historical wage patterns → new women workers receive lower predicted pay.

5. Opaque criteria

Workers cannot determine why their pay changed.

6. Unreliable performance metrics

Customer ratings or automated productivity measurements are inaccurate.

7. Algorithmic manipulation

The platform changes the variables or thresholds in ways workers cannot detect.

6. Case Law 1 — Defrenne v SABENA

Defrenne v Société Anonyme Belge de Navigation Aérienne

Case 43/75, CJEU, 8 April 1976

Defrenne is the foundational equal-pay case.

The CJEU held that the principle of equal pay contained in the Treaty had direct effect, meaning it could be invoked by an employee against a private employer. The CJEU itself identifies Defrenne as the foundational case concerning direct effect of the equal-pay principle. (curia)

Algorithmic relevance

Suppose:

Male workers and female workers perform equal work.

An employer cannot avoid Article 157 merely because:

"The wage was calculated by software."

The legal obligation attaches to the employment relationship and remuneration outcome, not to whether a human or algorithm performed the calculation.

Principle

Automation does not remove the fundamental right to equal pay.

Relevance: Very high.

7. Case Law 2 — Danfoss

Handels- og Kontorfunktionærernes Forbund i Danmark v Dansk Arbejdsgiverforening, acting for Danfoss

Case C-109/88, CJEU, 17 October 1989

Danfoss is particularly important for opaque wage systems.

The case concerned wage differences where the employer used various criteria in determining pay.

The CJEU addressed the evidential problem that workers may not have access to the information necessary to prove discrimination. The case is therefore a foundational authority concerning burden of proof where remuneration criteria are insufficiently transparent. (Infocuria)

Algorithmic relevance

Imagine:

AI determines individual wage adjustments.

Workers observe:

Worker A = €20/hour
Worker B = €17/hour

but cannot discover:

the scoring criteria;

weighting;

performance adjustments;

bonus calculations.

Danfoss becomes highly relevant because the employer's control over wage information should not make an equal-pay claim practically impossible.

Principle

Lack of transparency in remuneration systems can have important consequences for the allocation of the evidential burden.

Relevance: Extremely high.

8. Case Law 3 — Enderby

Enderby v Frenchay Health Authority

Case C-127/92, CJEU, 27 October 1993

Enderby is one of the most important cases for statistical evidence and indirect pay discrimination.

The case involved predominantly female and predominantly male occupations of equal value with significant differences in remuneration.

The CJEU held that significant statistics can establish a prima facie case, after which the employer must provide objective justification unrelated to sex. (curia)

Algorithmic relevance

Suppose an employer's algorithm produces:

Male workers: average €25/hour
Female workers: average €20/hour

The difference alone does not automatically establish unlawful discrimination.

But if the statistics reveal a significant disparity, they can become important evidence.

The employer may then have to explain:

what criteria generated the disparity;

whether those criteria are objective;

whether they are applied consistently;

whether they are gender-neutral.

Principle

Statistical disparities can shift the evidential burden where they establish a prima facie case of sex discrimination.

Relevance: Extremely high.

9. Case Law 4 — Royal Copenhagen

Royal Copenhagen A/S v Dansk Arbejderforbund

Case C-400/93, CJEU, 31 May 1995

Royal Copenhagen concerned remuneration involving variable elements and performance-related pay.

The case is important for understanding how differences in remuneration can be assessed where pay is affected by performance or productivity.

Algorithmic relevance

Modern algorithms frequently determine remuneration according to:

productivity;

output;

individual performance;

efficiency;

quality scores.

The employer can legitimately use performance-related criteria, but the relevant question is whether those criteria are applied objectively and without sex discrimination.

Principle

Performance-based remuneration is not inherently unlawful, but its criteria and application must remain compatible with equal-pay principles.

The CJEU later referred to Royal Copenhagen when explaining that assessment of whether work has equal value is ultimately a factual matter concerning the actual work performed. (Infocuria)

Relevance: High.

10. Case Law 5 — Brunnhofer

Susanna Brunnhofer v Bank der österreichischen Postsparkasse AG

Case C-381/99, CJEU, 26 June 2001

Brunnhofer is extremely useful for algorithmic wage-setting because it concerns:

unequal pay;

objective justification;

individual performance;

criteria used to distinguish wages.

The CJEU explained that differences in remuneration can potentially be justified by objective factors unrelated to sex, but such factors must genuinely justify the difference and comply with proportionality. (Infocuria)

Importantly, the Court addressed the limits of relying on differences in individual performance that become apparent only after employment begins. (curia)

Algorithmic relevance

Suppose:

AI performance score → lower wage.

The employer cannot simply say:

"The algorithm says Worker B is less effective."

It may have to explain:

what "effectiveness" means;

what data were used;

whether the data are reliable;

whether the metric is relevant;

whether the criterion is gender-neutral;

whether it was applied consistently.

Principle

An algorithmic performance criterion must be capable of constituting a genuine objective justification rather than merely being an unexplained computational output.

Relevance: Extremely high.

11. Case Law 6 — JämO

Jämställdhetsombudsmannen v Örebro läns landsting

Case C-236/98, CJEU, 30 March 2000

JämO concerned equal pay and the comparison of different occupations, including questions about supplements and inconvenient working hours.

The CJEU examined how remuneration components should be considered when comparing workers and how the burden of proof operates where apparent discrimination exists. (Infocuria)

Algorithmic relevance

Algorithms increasingly calculate remuneration using multiple components:

base pay + night premium + productivity bonus + customer rating + dynamic incentive.

The employer cannot necessarily isolate one component and ignore the broader remuneration structure.

Principle

Equal-pay analysis can require examination of the actual components of remuneration rather than merely comparing headline salaries.

Relevance: High.

12. Case Law 7 — Tesco Stores

Tesco Stores Ltd v USDAW

Case C-624/19, CJEU, 3 June 2021

Tesco is a landmark modern equal-pay decision.

The CJEU confirmed that Article 157 TFEU can be relied upon directly in disputes between private parties and applies to work of equal value, not merely identical work. (curia)

Algorithmic relevance

Suppose:

predominantly female retail workers → algorithmically lower pay

and:

predominantly male warehouse workers → higher pay.

The employer cannot necessarily defeat the claim simply by saying:

"They have different job titles."

The factual value of the work must be assessed.

Principle

Different jobs can still be compared for equal-pay purposes where they are of equal value.

Relevance: Extremely high.

13. Case Law 8 — Uber Spain

Asociación Profesional Elite Taxi v Uber Systems Spain SL

Case C-434/15, CJEU Grand Chamber, 20 December 2017

Uber was not an equal-pay case, but it is highly relevant to algorithmic management.

The CJEU recognised that Uber exercised significant control over important aspects of the service, including the organisation of the service through its platform.

Algorithmic wage relevance

Platform businesses may argue:

"The worker is paid according to an algorithm, not by an employer."

Uber demonstrates why courts examine the actual economic and organisational relationship, rather than relying exclusively on the technological description.

The algorithm may determine:

fares;

incentives;

allocation;

working opportunities;

performance evaluation.

Therefore, the question of who legally controls the work remains crucial.

Relevance: Analogical but important.

14. Case Law 9 — Yodel

B v Yodel Delivery Network Ltd

Case C-692/19, CJEU, 22 April 2020

Yodel concerned the status of a parcel courier and the characteristics of the working relationship.

Although it was not an algorithmic wage-discrimination case, it is relevant where algorithmic remuneration systems are used in delivery and platform work.

Relevance

Before determining whether a worker has an employment-law wage claim, the court may need to determine:

What is the legal status of the worker?

A platform cannot necessarily avoid employment protection merely by describing someone as an independent contractor.

Relevance: Analogical/high for platform wage disputes.

15. Case Law 10 — FNV Kunsten

FNV Kunsten Informatie en Media v Staat der Nederlanden

Case C-413/13, CJEU, 4 December 2014

The CJEU addressed the position of self-employed workers who may be in a situation comparable to employees.

The case is important for distinguishing genuine self-employment from situations where workers are economically dependent.

Algorithmic relevance

An organisation may say:

"Our algorithm pays independent contractors, so employment equality rules do not apply."

That statement cannot automatically resolve the issue.

The legal status depends upon the actual relationship.

Relevance: High by analogy.

16. New EU Pay Transparency Rules

Directive 2023/970 is especially significant for future algorithmic wage litigation.

Article 6

Employers must make accessible the criteria used to determine:

pay;

pay levels;

pay progression.

The criteria must be:

objective;

gender-neutral. (EUR-Lex)

This has major implications for algorithmic systems.

An employer using:

AI → performance score → pay increase

will increasingly need to be able to identify the criteria used for the remuneration process.

17. Right to Pay Information

Article 7 of Directive 2023/970 gives workers rights to request:

their individual pay level;

average pay levels;

sex-disaggregated information for categories of workers performing the same work or work of equal value.

Workers can request additional reasonable clarification where information is incomplete or inaccurate. (EUR-Lex)

This is particularly important because algorithmic wage systems can create an information asymmetry:

Employer/platform knows the algorithm.
Worker sees only the final amount.

Pay transparency law attempts to reduce that asymmetry.

18. Algorithms and Gender-Neutral Criteria

A company may argue:

"The algorithm does not know whether the worker is male or female."

That is not necessarily sufficient.

The relevant question can be whether the algorithm uses indirect proxies or criteria that produce discriminatory effects.

For example:

availability at particular times → higher algorithmic bonus

may disproportionately affect workers with certain family or caring patterns.

Another example:

historical earnings → predicted future wage

could reproduce historical gender disparities.

Therefore:

No explicit gender variable ≠ automatically no gender discrimination.

19. Historical Wage Bias

This is one of the most serious algorithmic problems.

Suppose historical data show:

Men historically earned €30/hour
Women historically earned €24/hour.

An algorithm learns from that dataset.

It predicts:

male worker → higher expected productivity → higher pay

and:

female worker → lower expected productivity → lower pay.

The algorithm has transformed historical inequality into a predictive wage rule.

This can create:

PAST DISCRIMINATION → TRAINING DATA → AI MODEL → NEW PAY DIFFERENCE → NEW DATA → REINFORCED DISCRIMINATION

20. Performance-Based Algorithms

Performance-based pay is not inherently unlawful.

Employers can generally use objective criteria such as:

experience;

skills;

productivity;

responsibility;

working conditions.

Indeed, Directive 2023/970 expressly contemplates objective, gender-neutral, bias-free criteria such as performance and competence. (EUR-Lex)

The legal difficulty arises when the "performance" measure itself is:

biased;

inaccurate;

opaque;

selectively applied;

based on discriminatory proxies.

21. Customer Ratings as Wage Criteria

Consider a delivery platform.

Worker A receives:

4.9/5 rating

Worker B receives:

4.5/5 rating.

The algorithm reduces Worker B's incentive payments.

Potential questions include:

Are customer ratings accurate?

Do customers rate different groups differently?

Does language affect ratings?

Does geographic location affect ratings?

Are ratings adjusted for bias?

Can the worker challenge inaccurate ratings?

Is the rating criterion transparent?

Does it disproportionately disadvantage a protected group?

An automated score is therefore evidence, not automatically a legally valid justification.

22. Dynamic Wage Algorithms

Dynamic algorithms can change remuneration according to:

demand;

location;

time;

weather;

worker availability;

customer demand;

predicted acceptance rates.

Example:

10:00 AM → €10/hour
7:00 PM → €18/hour

This may be legitimate.

But the system becomes legally problematic if the differential treatment is connected to prohibited discrimination or violates contractual/statutory wage requirements.

The key issue is:

OBJECTIVE CRITERION → CONSISTENT APPLICATION → NON-DISCRIMINATION → TRANSPARENCY

23. Platform Work Directive and Earnings

Directive 2024/2831 expressly recognises that automated systems may affect platform workers' earnings. (EUR-Lex)

The Directive's transparency rules cover automated systems affecting working conditions, including:

work assignments;

earnings;

working time;

performance;

promotion;

contractual status.

(EUR-Lex)

This is particularly important because algorithmic wage setting is often embedded inside a larger algorithmic-management system.

24. Human Oversight

Article 10 of Directive 2024/2831 requires effective human oversight and evaluation of automated decisions affecting platform workers.

The people performing the oversight must have:

competence;

training;

authority;

ability to override automated decisions.

(EUR-Lex)

This is important where:

AI calculates worker remuneration → worker disputes result.

A genuine human review mechanism can provide an important safeguard.

25. Human Review Does Not Automatically Cure Discrimination

Suppose:

Algorithm produces lower wage → manager clicks "approve."

That does not automatically answer the legal question.

The court may need to ask:

Did the manager actually review the data?

Could the manager override the system?

Did the manager understand the relevant criteria?

Was the worker allowed to challenge the result?

Did the manager independently verify the information?

Therefore:

FORMAL HUMAN APPROVAL ≠ NECESSARILY MEANINGFUL HUMAN REVIEW

26. Algorithmic Wage Discrimination and Burden of Proof

A typical litigation sequence could be:

Stage 1

Worker establishes:

Male comparator earns €25/hour.

Stage 2

Worker establishes:

Female worker performs equal or equal-value work but earns €20/hour.

Stage 3

Statistical evidence shows:

algorithm systematically pays male workers more.

Stage 4

The employer must provide objective justification, depending on the applicable equality framework.

Stage 5

The worker challenges the algorithmic criterion.

Stage 6

Court examines:

Is the criterion genuine, objective, proportionate and non-discriminatory?

This is where Danfoss, Enderby and Brunnhofer become particularly valuable. (Infocuria)

27. Indirect Discrimination

Indirect discrimination is particularly relevant to AI.

Example:

Algorithm gives substantial pay bonuses for workers available during evenings and weekends.

This criterion appears neutral.

But if it places women at a particular disadvantage because of statistically unequal caring responsibilities, the court may need to examine:

whether there is a particular disadvantage;

whether the criterion is objectively justified;

whether the measure is appropriate and necessary.

Thus:

NEUTRAL CRITERION → GROUP DISADVANTAGE → OBJECTIVE JUSTIFICATION → PROPORTIONALITY

28. Equal Value of Work

Tesco is especially important.

The comparison is not necessarily:

identical job title.

The question is whether the actual work has equal value.

Relevant factors can include:

skills;

effort;

responsibility;

working conditions.

Directive 2023/970 expressly requires pay structures to permit comparison using objective, gender-neutral criteria and identifies skills, effort, responsibility and working conditions as relevant factors. (EUR-Lex)

This is important for AI because a model may classify jobs differently even though they have equivalent value.

29. Algorithmic Job Evaluation

Imagine AI evaluates two occupations:

Occupation A

Predominantly male:

high responsibility = 90 points.

Occupation B

Predominantly female:

high emotional/social skills = 55 points.

If the algorithm undervalues skills associated with female-dominated occupations, the system may reproduce structural pay inequality.

Directive 2023/970 expressly recognises the importance of gender-neutral job evaluation and warns against undervaluing relevant soft skills. (EUR-Lex)

30. Civil Liability

A worker's claim may potentially involve:

A. Equal-pay claim

Wage difference violates equal-pay principles.

B. Discrimination claim

Algorithm causes direct or indirect discrimination.

C. Contractual claim

Employer failed to pay contractual remuneration.

D. Data-protection claim

Personal data were unlawfully processed or automated decision safeguards were breached.

E. Damages

Worker suffered financial or non-material loss.

F. Injunctive/declaratory relief

Algorithm or wage practice must be modified or discontinued.

The exact remedies depend on national law implementing EU obligations.

31. Causation

The claimant must often connect the algorithm to the wage loss.

For example:

Worker data

↓

Algorithmic score

↓

Lower performance classification

↓

Lower bonus

↓

Lower monthly remuneration

↓

Financial loss

This is a clearer causal chain than simply showing that:

"The company uses AI."

The use of AI itself does not create liability.

32. Evidence

Important evidence may include:

Algorithmic evidence

model documentation;

scoring rules;

variables;

weighting;

thresholds;

model versions;

change logs.

Employment evidence

contracts;

payslips;

bonus records;

performance reviews;

job descriptions.

Statistical evidence

male/female pay distributions;

bonus distributions;

promotion rates;

algorithmic scores;

error rates.

Technical evidence

audit reports;

bias testing;

validation;

monitoring records.

Comparative evidence

comparator workers;

job classifications;

workload;

responsibilities;

skills.

33. Trade Secrets

An employer may argue:

"The wage algorithm is commercially confidential."

That does not automatically eliminate an employee's statutory rights.

The legal system must balance:

trade secrets;

intellectual property;

privacy;

cybersecurity;

against:

equal-pay enforcement;

effective judicial protection;

transparency;

ability to prove discrimination.

The practical issue is therefore often not:

"Give the worker all source code."

but:

"Give enough legally relevant information to make the worker's rights effective."

34. Algorithmic Wage Setting and GDPR

Where worker personal data are used, GDPR may become relevant.

For example:

productivity data → profiling → wage determination.

Questions include:

What data are collected?

Is the processing lawful?

Is it necessary?

Are data accurate?

Is profiling occurring?

Does Article 22 apply?

Does the worker have relevant access rights?

Can inaccurate data be corrected?

Where an automated system has significant effects, the reasoning in SCHUFA concerning automated scoring can become relevant by analogy.

35. Algorithmic Wage Setting and AI Act

The AI Act can also become relevant depending on the particular system and its classification.

Employment-related AI systems can fall within the AI Act's high-risk framework in specified circumstances.

However, it is important not to say that:

"Every payroll algorithm is automatically high-risk AI."

That would be too broad.

The actual purpose and classification of the system must be examined.

36. Platform Workers and Employment Status

Before bringing an equal-pay employment claim, the court may need to determine:

Is this person legally a worker?

This is important in platform disputes.

Relevant CJEU authorities include:

Uber, C-434/15

Yodel, C-692/19

FNV Kunsten, C-413/13

These cases demonstrate the importance of examining the actual economic relationship, rather than relying only on contractual labels.

37. Example: AI Delivery-Wage System

Assume a delivery platform uses:

AI → customer rating → predicted reliability → bonus.

It discovers:

Male workers average €22/hour.
Female workers average €17/hour.

A female worker challenges the system.

Step 1 — Comparator

Are the workers performing equal or equal-value work?

Tesco / Enderby / Brunnhofer

↓

Step 2 — Statistical disparity

Is the difference significant?

Enderby

↓

Step 3 — Algorithmic criterion

Why does the algorithm produce the difference?

Danfoss / Brunnhofer

↓

Step 4 — Objective justification

Does the platform have a legitimate, objective and non-discriminatory explanation?

↓

Step 5 — Transparency

Can the worker obtain information about pay-setting criteria?

Directive 2023/970

↓

Step 6 — Algorithmic management

Is the worker covered by platform-work protections?

Directive 2024/2831

↓

Step 7 — Human review

Can a trained human override the automated decision?

Directive 2024/2831

↓

Step 8 — Remedy

Back pay + compensation + correction of algorithmic process, depending on applicable national law.

38. Important Distinction: Unfairness vs Illegality

An algorithm may be:

unfair-looking

without necessarily being:

legally discriminatory.

For example:

Worker A receives a higher bonus because Worker A sells twice as much.

That could be perfectly legitimate if the criterion is objectively relevant and applied consistently.

The stronger legal claim is:

The algorithm creates a prohibited difference in treatment, or violates a specific statutory/contractual obligation.

Therefore, legal analysis should not simply equate:

different pay = unlawful discrimination.

39. Key Case-Law Table

40. Direct vs Analogical Case Law

Because this is a relatively new field, accuracy requires distinguishing the cases.

Direct equal-pay authorities

Defrenne
Equal-pay principle.

Danfoss
Opaque wage criteria and evidential burden.

Enderby
Statistical discrimination.

Brunnhofer
Objective justification.

Tesco
Work of equal value.

JämO
Remuneration components.

Algorithm/platform analogies

Uber

Yodel

FNV Kunsten

These do not establish that an algorithmic wage system is discriminatory. They provide principles concerning platform control, worker status and economic dependence.

Modern statutory development

Directive 2023/970 is particularly important because it expressly requires transparency concerning criteria determining pay and pay progression. (EUR-Lex)

Directive 2024/2831 is particularly important for platform algorithmic management, including systems affecting earnings and human oversight. (EUR-Lex)

41. Strongest Legal Argument

The strongest algorithmic wage-discrimination claim generally looks like:

The worker performs equal or equal-value work → the algorithm produces a material pay disparity → statistical evidence demonstrates a group disadvantage → the algorithm uses criteria that are not sufficiently objective/gender-neutral → the employer cannot adequately justify the difference → the worker seeks the relevant information → the disparity and algorithmic mechanism are established → compensation and other remedies are sought under applicable law.

42. Main Defences by Employers

Employers may argue:

1. Genuine performance difference

The worker genuinely produced different results.

2. Objective criteria

The algorithm uses neutral factors.

3. Different work

The employees are not performing equal or equal-value work.

4. Statistical coincidence

The observed disparity is temporary or random.

5. Worker-specific factors

Experience, skills or responsibility explain the difference.

6. Human decision

The algorithm only assists a human manager.

7. No employment relationship

The claimant is an independent contractor.

8. Confidentiality

Algorithmic details constitute trade secrets.

The court must examine the evidence rather than accept the label "AI" or "objective algorithm" at face value.

43. Future Litigation Trend

The combination of:

Article 157 TFEU + Directive 2006/54 + Pay Transparency Directive 2023/970 + Platform Work Directive 2024/2831 + GDPR + national employment law

is likely to make algorithmic remuneration increasingly litigated.

The most significant change is that transparency itself is becoming more enforceable.

The Pay Transparency Directive expressly identifies lack of information about pay structures as an obstacle to equal-pay litigation and seeks to enable workers to identify discrimination. (EUR-Lex)

44. Ultra-Basic Exam Notes

Meaning

Algorithmic wage-setting fairness litigation = legal challenge to wages, bonuses or incentives calculated or influenced by AI/automated systems.

Main problems

Gender discrimination

Indirect discrimination

Proxy discrimination

Biased historical data

Opaque performance scoring

Inaccurate worker data

Unfair bonuses

Unequal job valuation

Lack of pay transparency

Platform-worker exploitation

Main laws

Article 157 TFEU

Directive 2006/54/EC

Directive 2023/970

Directive 2024/2831

GDPR

AI Act

National employment and equality law

Key cases

Defrenne C-43/75 → equal pay has direct effect

Danfoss C-109/88 → opaque wage criteria/evidential burden

Enderby C-127/92 → statistics and burden of proof

Royal Copenhagen C-400/93 → performance-related pay

Brunnhofer C-381/99 → objective justification

JämO C-236/98 → remuneration components

Tesco C-624/19 → equal value of work

Uber C-434/15 → platform control

Yodel C-692/19 → worker status

FNV Kunsten C-413/13 → false self-employment

45. Master Legal Formula

WORKER DATA → ALGORITHM → PERFORMANCE SCORE → WAGE/BONUS → PAY DIFFERENCE → COMPARATOR → EQUAL/ EQUAL-VALUE WORK → STATISTICAL EVIDENCE → OBJECTIVE JUSTIFICATION → TRANSPARENCY → DISCRIMINATION TEST → CAUSATION → DAMAGES/REMEDY

Shortest exam formula:

AI → PAY → DIFFERENCE → EQUALITY → TRANSPARENCY → JUSTIFICATION → CAUSATION → REMEDY

Final conclusion

European law does not prohibit algorithmic wage-setting as such. The legal problem arises when automated remuneration produces unjustified discriminatory differences, uses non-objective or biased criteria, prevents effective equal-pay enforcement, or operates without the transparency and human safeguards required by applicable EU and national law. The combination of Danfoss, Enderby, Brunnhofer and Tesco provides the established equal-pay foundation, while Directives 2023/970 and 2024/2831 provide an increasingly explicit framework for transparency and algorithmic management affecting remuneration. (Infocuria)