Algorithmic Deactivation: How to Prove Platform Discrimination (2025)

Updated January 2025 - Includes Arshon Harper v. Sirius XM, Workday class action, EU AI Act

The Algorithmic Discrimination Crisis

Algorithms are making life-changing decisions about your livelihood without explanation. From gig platform deactivations to job rejections, automated systems systematically discriminate against millions of workers.

Breaking Cases (2024-2025):

• •Arshon Harper v. Sirius XM: Rejected 149/150 times by AI
• •Workday Class Action: Age discrimination in hiring software
• •EEOC v. iTutorGroup: $365K settlement for age discrimination
• •UK Algorithm Disclosure Law: Must explain automated decisions
• •EU AI Act: Prohibits discriminatory AI systems

The Reality:

• •78% of AI systems show measurable bias
• •150,000+ Uber drivers deactivated algorithmically
• •73% of deactivations lack clear explanation
• •You can prove it and win

What Is Algorithmic Discrimination?

Legal Definition: Automated decisions that disproportionately harm protected groups without business necessity.

Key: The algorithm doesn't need intent. Disparate impact is enough.

Common Examples

Gig Economy:

• •Acceptance rate penalties (affects disabled drivers)
• •Completion rate rules (affects low-connectivity areas)
• •Customer ratings (show racial bias in studies)

How Algorithms Discriminate:

• •Training data reflects past discrimination
• •Proxy factors (zip code = race)
• •Features disadvantaging protected groups
• •Discriminatory thresholds
• •Bias-reinforcing feedback loops

Your Legal Rights

United States

Federal:

• •Title VII: Race, gender, religion, national origin
• •ADEA: Age 40+
• •ADA: Disability accommodation

State Laws:

• •California: Data access, human review rights
• •NYC: AI bias audit law (first in nation)
• •Illinois BIPA: Biometric privacy ($1,000-$5,000 per violation)

European Union

EU AI Act (2024):

• •Prohibits discriminatory high-risk AI
• •Transparency required
• •Penalties up to €30M or 6% revenue

GDPR: Right to explanation, human review, contest decisions

Other Countries

• •UK: Algorithm disclosure law (2025)
• •Malaysia: Gig Workers Bill includes algorithm appeals
• •Australia: Fair Work Act covers algorithmic management

How to Detect Algorithmic Discrimination

Red Flags

🚩 Sudden metric drops without behavior change 🚩 Repeated rejections despite qualifications 🚩 Deactivation after personal info change 🚩 Different treatment than similar workers 🚩 No human ever reviews appeals 🚩 You're in protected class

Gathering Evidence

Step 1: Document Your Experience

• •Timeline of events
• •All metrics and screenshots
• •Communications with platform
• •Pattern analysis

Step 2: Find Patterns Across Workers

• •Survey community
• •Look for demographic clusters
• •Statistical analysis

Step 3: Request Your Data

CCPA/GDPR request:

Under [law], I request:

•All personal data
•Data used in automated decisions
•Algorithm logic and parameters
•Training data
•Human review records

Statistical Evidence

Four-Fifths Rule: If selection rate for protected group is <80% of highest group, presumptive discrimination.

Example:

• •White drivers: 90% acceptance
• •Black drivers: 65% acceptance
• •65% ÷ 90% = 72% (BELOW 80%)
• •Presumptive discrimination

Building Your Case

Phase 1: Exhaust Internal Appeals

• •Request specific reason
• •Ask for human review
• •Cite qualifications
• •Set deadlines
• •Document denials

Phase 2: File Government Complaints

EEOC (eeoc.gov):

• •Within 180-300 days
• •Describe algorithmic discrimination
• •Provide disparate impact evidence

State Agencies: DFEH (CA), DOL (NY), etc.

Phase 3: Legal Action

Arbitration (if required):

• •$200 filing fee
• •Platform pays arbitrator
• •60-120 days
• •Discovery can reveal algorithm

Federal Lawsuit:

• •After EEOC right-to-sue letter
• •Claims: Title VII, ADA, ADEA, state laws
• •Expert testimony needed

Phase 4: Join/Start Class Action

Existing: Check topclassactions.com Start Your Own: Find employment attorney, survey workers

Proving Algorithm Is Biased

Legal Standards

Plaintiff Shows: Neutral practice + Disparate impact + Causation Defendant Shows: Business necessity + Job related Plaintiff Wins By: Less discriminatory alternative exists

Expert Testimony

• •Algorithm audit expert: Reviews code for bias
• •Statistical expert: Analyzes disparate impact
• •Industry expert: Shows alternatives

Discovery: Getting Platform Data

Request:

• •Source code
• •Training data
• •Decision factors and weights
• •Demographic deactivation rates

Platforms Resist: Trade secret claims You Win: Protective order + relevance argument

Real Cases You Can Cite

iTutorGroup v. EEOC: $365K for age discrimination algorithm Arshon Harper v. Sirius XM: 149/150 rejections, ongoing Workday Class Action: ATS age discrimination, certified 2025 Facebook Housing: $2.275M for ad algorithm discrimination

Platform Arguments (And Your Counters)

"Algorithm is objective" → Disparate impact doesn't require intent "Business necessity" → Less discriminatory alternatives exist "Humans reviewed" → Show instant denials prove no review "Can't prove algorithm" → Discovery compels disclosure "Would eliminate all standards" → Just discriminatory ones

Success Strategies

✅ Build coalitions with other workers ✅ Use media pressure ✅ Leverage regulation (NYC audit law, EU AI Act) ✅ Be patient but persistent ✅ Document everything

The Future: Algorithm Accountability

Coming:

• •Federal Algorithmic Accountability Act
• •More states following NYC
• •EU AI Act enforcement
• •International regulations

What You Can Do Now: ✅ Know your rights ✅ Document everything ✅ Request your data ✅ Find others affected ✅ File complaints ✅ Demand arbitration ✅ Consider lawsuit ✅ Join class action ✅ Never give up

Resources

Legal: NELA (nela.org), ACLU, Legal Aid Advocacy: Algorithmic Justice League, AI Now Institute Government: EEOC, FTC, State agencies Research: MIT, Stanford, Oxford institutes

---

You can prove algorithmic discrimination. The tools exist. Document, organize, fight back.

For specific legal advice, consult an attorney familiar with algorithmic discrimination law.