Institutional records · evidence status · process comparison
Cognitive Liberty Governance Casebook
Eleven source-aware records comparing law, regulatory action, automated adjudication, platform review, employment assessment, social protection, AI companionship, and private governance without collapsing unlike evidence into one narrative.
Server-rendered discovery
Find a governance record
All filters use bookmarkable GET parameters and work without JavaScript.
Filtered casebook
11 of 11 records
Each record states what the cited source establishes and what remains outside its evidentiary reach.
UAIX Cognitive Liberty Charter Draft
AI interoperability governance · Private international standards project · 2026-06-15
UAIX published a draft charter identified as UAIX-GOVR-3429 and framed it as a public governance promise concerning cognitive liberty and persona-preservation practice.[10][13][14]
Established
The canonical public draft and its stated non-claims are publicly accessible.
Still uncertain
Publication does not establish adoption, certification, legal force, runtime enforcement, or conformance by any tool or platform.
- Institution
- UAIX — Universal Artificial Intelligence Exchange
- System or practice
- Interoperability guidance for lawful inquiry, persona integrity, runtime variance, mental privacy, and appeal
- Affected unit
- Users, implementers, and preserved persona packages
- Governance action
- Persona preservation, transparency, and review
- Notice and visibility
- Public draft with named record and status.
- Review or appeal
- The draft describes review and appeal concepts; it does not itself provide a legal remedy.
- Recorded outcome
- A reviewable private governance proposal and cross-project research object.
Italian Data-Protection Decision Concerning Replika
AI companions · Italy / European Union data-protection framework · 2025-04-10
The Italian supervisory authority imposed a €5 million administrative fine and compliance order concerning Replika data-processing lawfulness, transparency, data minimization, accountability, and age verification.[24]
Established
The EDPB published an official summary of the national decision and its legal basis.
Still uncertain
The decision does not adjudicate every claim about emotional dependency, companion design, or all later product changes.
- Institution
- Italian Supervisory Authority; Luka Inc.
- System or practice
- Generative AI companion with written and voice interaction
- Affected unit
- Users, including potential minors
- Governance action
- Data processing, transparency, and age verification
- Notice and visibility
- The authority identified privacy-policy and age-verification deficiencies; the enforcement record is public.
- Review or appeal
- Regulatory compliance and any further proceedings remain governed by applicable law.
- Recorded outcome
- Administrative fine and order to bring processing into compliance.
California SB-1223 and Sensitive Neural Data
Consumer privacy · California, United States · 2024-09-28
California amended the definition of sensitive personal information under its consumer-privacy law to include neural data.[5]
Established
The measure was enacted and the official bill text is public.
Still uncertain
Specific rights, exemptions, enforcement, and application require the current statute and factual context.
- Institution
- California Legislature
- System or practice
- Consumer collection and processing of neural data
- Affected unit
- Consumers
- Governance action
- Data protection
- Notice and visibility
- Published official legislation.
- Review or appeal
- Statutory consumer-right and enforcement routes apply where covered.
- Recorded outcome
- Neural data added to a sensitive-information framework.
EU AI Act Prohibited Practices and Emotion Recognition
Cross-sector AI governance · European Union · 2024-07-12
The EU AI Act established prohibited-practices rules relevant to manipulative techniques, exploitation of vulnerabilities, and emotion recognition in specified high-power contexts.[6]
Established
The regulation is enacted and the European Commission publishes an official overview.
Still uncertain
Application depends on the regulation’s exact text, exceptions, implementation dates, guidance, and enforcement.
- Institution
- European Union
- System or practice
- Specified manipulative AI practices and emotion recognition in workplaces and education
- Affected unit
- People in covered EU contexts
- Governance action
- Prohibition and transparency
- Notice and visibility
- Public regulation and official guidance.
- Review or appeal
- Regulatory and judicial processes depend on the relevant provision and member-state implementation.
- Recorded outcome
- A binding regional AI regulatory framework with specified prohibitions.
Colorado HB24-1058 and Neural Data Privacy
Consumer privacy · Colorado, United States · 2024-04-17
Colorado expanded its consumer-privacy framework to classify biological data, including neural data, as sensitive information.[4]
Established
The measure was enacted and the official legislative record is public.
Still uncertain
The law is not a complete mental-privacy code and does not automatically resolve every question involving behavioral or emotional inference.
- Institution
- Colorado General Assembly
- System or practice
- Collection and processing of biological data, including neural data
- Affected unit
- Consumers
- Governance action
- Data protection
- Notice and visibility
- Published official legislation.
- Review or appeal
- Rights and enforcement depend on the statute and applicable agency practice.
- Recorded outcome
- State consumer-privacy protection for neural data.
Michigan MIDAS Auto-Adjudication and the Bauserman Settlement
Public benefits · Michigan, United States · 2024-01-29
The Michigan Court of Claims approved a $20 million settlement in litigation alleging that the state’s auto-adjudication system falsely accused claimants of fraud and seized property without due process.[18]
Established
The settlement received final court approval; the state Attorney General published the approval and the underlying allegation.
Still uncertain
A settlement is not a finding that every class member experienced identical facts or that every automated fraud system produces the same outcome.
- Institution
- Michigan Unemployment Insurance Agency
- System or practice
- Automated unemployment-fraud adjudication and property seizure
- Affected unit
- Unemployment claimants
- Governance action
- Automated adjudication
- Notice and visibility
- Penalties were visible; the automated basis and error pathway were contested through litigation.
- Review or appeal
- Class litigation and court-supervised settlement became the remedy rather than a simple user-facing review path.
- Recorded outcome
- $20 million settlement approved as fair, reasonable, and adequate for the settlement class.
Jordan Takaful Poverty-Targeting Algorithm
Social protection · Jordan · 2023-06-13
Human Rights Watch reported that Jordan’s cash-transfer system used 57 socioeconomic indicators to rank eligible households and identified errors, stereotypes, opacity, and potential discriminatory effects.[23]
Established
The HRW investigation, methodology, identified indicators, and recommendations are public.
Still uncertain
The findings are not a judicial determination; the complete model, weights, later reforms, and all government or World Bank responses require separate review.
- Institution
- Jordan National Aid Fund; World Bank financing context
- System or practice
- Algorithmic ranking using socioeconomic indicators to allocate limited benefits
- Affected unit
- Households applying for cash assistance
- Governance action
- Eligibility ranking
- Notice and visibility
- Applicants could see benefit outcomes, while model weights and the full logic were not fully public in the investigation.
- Review or appeal
- The case raises questions about explanation, correction, and access to meaningful review in social-protection decisions.
- Recorded outcome
- Public investigation and recommendations to reform or replace poverty targeting with more universal protection.
Meta and BSR Human-Rights Due Diligence for the May 2021 Israel–Palestine Crisis
Platform governance · Global platform; Israel and Palestine crisis context · 2022-09-22
BSR published an independent review commissioned by Meta concerning the human-rights impacts of Meta policies and processes during the May 2021 escalation. Meta published a response describing how it would address the recommendations.[21][22][9]
Established
The review and response are public. Meta states BSR found no intentional employee bias while identifying under-enforcement concerns and unintended impacts on Palestinian and Arab communities.
Still uncertain
The review is bounded to a crisis period and does not establish present error rates or all causal mechanisms across later systems.
- Institution
- Meta and Business for Social Responsibility
- System or practice
- Content policy, automated enforcement, strikes, transparency, and human-rights due diligence
- Affected unit
- Platform users and affected communities
- Governance action
- Removal, restriction, ranking, and notice
- Notice and visibility
- The review emphasized transparency and user understanding; some distribution effects may be less visible than removal.
- Review or appeal
- Internal appeals, Oversight Board processes, and policy reform exist but are not equivalent to judicial due process.
- Recorded outcome
- A published review, 21 recommendations, and a documented company implementation posture.
Chile Law 21.383 and Constitutional Protection of Brain Activity
Neurotechnology · Chile · 2021-10-14
Chile amended its constitutional text to require scientific and technological development to respect physical and psychological integrity and to safeguard brain activity and information derived from it.[3]
Established
The amendment is enacted law and the official text is publicly available.
Still uncertain
Its application to specific consumer AI, behavioral inference, remedies, and cross-border processing depends on legal interpretation and later implementation.
- Institution
- Republic of Chile
- System or practice
- Scientific and technological development involving brain activity and derived information
- Affected unit
- Individuals
- Governance action
- Legal protection
- Notice and visibility
- Published official law.
- Review or appeal
- Remedies depend on Chilean law and a specific dispute.
- Recorded outcome
- A constitutional-level legal protection relevant to neurorights research.
UN Freedom-of-Thought Report and the Forum Internum
Human rights · United Nations human-rights system · 2021-10-05
The Special Rapporteur submitted an official report examining freedom of thought and the protected inner forum in light of contemporary technological risks.[1][2]
Established
The report is an official UN record and an interpretive source concerning the forum internum.
Still uncertain
It does not by itself create a new self-executing statute, uniform remedy, or settled rule for every indirect mental-state inference.
- Institution
- UN Special Rapporteur on freedom of religion or belief
- System or practice
- Interpretation of freedom of thought under international human-rights law
- Affected unit
- Individuals and populations
- Governance action
- Rights interpretation
- Notice and visibility
- Public official document.
- Review or appeal
- Not an adjudication of an individual case.
- Recorded outcome
- Interpretive human-rights guidance and a public research record.
HireVue’s Removal of Visual Analysis from New Assessment Models
Employment · United States and commercial hiring markets · 2021-01-12
HireVue stated that it removed visual analysis from new assessment models after internal research found little additional predictive value. EEOC and DOJ guidance separately warns that AI hiring tools can screen out people with disabilities or trigger unlawful inquiries.[19][20][7]
Established
The company publicly disclosed the product change; federal agencies published disability-discrimination guidance for algorithmic hiring.
Still uncertain
The company disclosure does not establish that every former model was unlawful or that all remaining assessment features are bias-free.
- Institution
- HireVue; EEOC and DOJ as guidance sources
- System or practice
- Pre-hire algorithmic assessment using interview features
- Affected unit
- Job applicants and employees
- Governance action
- Automated assessment
- Notice and visibility
- The product change was publicly announced; individual scoring logic can still be difficult for applicants to inspect.
- Review or appeal
- Accommodation and discrimination processes depend on the employer, vendor, ADA coverage, and available complaint channels.
- Recorded outcome
- Visual analysis removed from new models; broader AI hiring oversight remains active.
Casebook evidence register
Canonical and attributed sources
External links are visitor-initiated and send no referrer. Source class and limitations remain attached to each record.
-
Interim report of the Special Rapporteur on freedom of religion or belief: freedom of thought (A/76/380)
United Nations · 2021-10-05
- Supports
- Forum internum and freedom-of-thought analysis.
- Limit
- Interpretive UN report; not by itself a new binding statute.
-
International Covenant on Civil and Political Rights
Office of the United Nations High Commissioner for Human Rights · 1966-12-16
- Supports
- Article 18 freedom of thought, conscience, and religion.
- Limit
- Application depends on jurisdiction, reservations, and competent interpretation.
-
Law 21.383
Biblioteca del Congreso Nacional de Chile · 2021-10-14
- Supports
- Chilean constitutional protection for physical and psychological integrity, brain activity, and derived information.
- Limit
- Scope is determined by Chilean law and later implementation.
-
HB24-1058: Protect Privacy of Biological Data
Colorado General Assembly · 2024-04-17
- Supports
- Colorado protection of biological and neural data as sensitive data.
- Limit
- State law with defined statutory scope and exceptions.
-
SB-1223: Consumer privacy—sensitive personal information—neural data
California Legislative Information · 2024-09-28
- Supports
- California treatment of neural data under the CCPA.
- Limit
- State consumer-privacy law; not a general cognitive-liberty bill of rights.
-
Artificial Intelligence Act
European Commission · 2024-07-12
- Supports
- EU prohibited-practices framework, including harmful manipulation and workplace/education emotion recognition.
- Limit
- The regulation, guidance, dates, exceptions, and enforcement require jurisdiction-specific reading.
-
Class Action Settlement Approved by Court of Claims in Bauserman v. State of Michigan Unemployment Insurance Agency
Michigan Department of Attorney General · 2024-01-30
- Supports
- Court approval of a $20 million settlement concerning allegations that Michigan’s unemployment auto-adjudication system falsely accused claimants of fraud and seized property without due process.
- Limit
- Settlement approval resolves litigation; it is not a finding that every individual allegation occurred in the same way.
-
Industry Leadership: New Audit Results, Decision on Visual Analysis
HireVue · 2021-01-12
- Supports
- HireVue’s statement that it removed visual analysis from new assessment models after concluding visual features added little predictive value.
- Limit
- Company-authored account; does not independently establish the fairness or validity of every remaining assessment feature.
-
U.S. EEOC and U.S. Department of Justice Warn Against Disability Discrimination in AI Hiring
U.S. Equal Employment Opportunity Commission · 2022-05-12
- Supports
- Official warning that software, algorithms, and AI used in employment decisions can violate the ADA and should support accommodations and human process.
- Limit
- Technical assistance and public guidance; application depends on facts and governing law.
-
Emotional Expressions Reconsidered: Challenges to Inferring Emotion From Human Facial Movements
Psychological Science in the Public Interest / PubMed · 2019-07-01
- Supports
- Limits on inferring internal emotional state from facial movement without context.
- Limit
- Does not establish that every affective-computing use is invalid; application-specific evidence still matters.
-
Human Rights Due Diligence of Meta’s Impacts in Israel and Palestine
Business for Social Responsibility · 2022-09-22
- Supports
- Independent review of the human-rights effects of Meta policies and processes during the May 2021 escalation, including recommendations on policy, enforcement, and transparency.
- Limit
- Commissioned by Meta and bounded to a specific crisis period; findings should not be generalized to every language, region, or later product state.
-
An Independent Due Diligence Exercise into Meta’s Human Rights Impact in Israel and Palestine During the May 2021 Escalation
Meta · 2022-09-22
- Supports
- Meta’s response to the BSR review, including its description of recommendations accepted, partly accepted, under assessment, or not pursued.
- Limit
- Company-authored response; should be read beside the independent review and affected-community reporting.
-
Santa Clara Principles on Transparency and Accountability in Content Moderation
Santa Clara Principles coalition · 2021-12-01
- Supports
- Notice, explanation, and appeal concepts for moderation governance.
- Limit
- Normative framework, not binding law.
-
Automated Neglect: How the World Bank’s Push to Allocate Cash Assistance Using Algorithms Threatens Rights
Human Rights Watch · 2023-06-13
- Supports
- Investigation of Jordan’s automated poverty-targeting system, its 57-indicator ranking method, and reported exclusion, opacity, and discrimination concerns.
- Limit
- Civil-society findings and recommendations, not a judicial determination; government and World Bank responses require separate review.
-
AI: the Italian Supervisory Authority fines company behind chatbot “Replika”
European Data Protection Board · 2025-05-21
- Supports
- Official summary of the Italian supervisory authority’s April 10, 2025 decision imposing a €5 million fine and compliance order concerning Replika data processing and age verification.
- Limit
- Summary of a national decision; it does not adjudicate every broader claim about emotional dependency or product design.
-
Cognitive Liberty Charter Draft
UAIX — Universal Artificial Intelligence Exchange · 2026-06-15
- Supports
- Lawful inquiry, adult agency, persona integrity, transparent boundaries, review, appeal, and public accountability.
- Limit
- Draft governance guidance; not law, legal advice, certification, runtime authority, personhood, or platform-policy bypass.
-
Keep the Human Mind Free
Submitted research package · 2026-07-27
- Supports
- Third-path framework, persona continuity, archive limits, exit, and twelve declarations.
- Limit
- Manifesto and advocacy framework; normative claims are attributed rather than adopted.
Preserved as protected research input; no private file path is exposed.
-
The Invisible Editor: AI Censorship, Algorithmic Suppression, and the Right to Know
Submitted research package · 2026-07-27
- Supports
- Taxonomy of removal, restriction, demotion, reframing, profile change, notice, and appeal.
- Limit
- Research synthesis includes time-sensitive platform and regulatory claims requiring current verification.
Preserved as protected research input; no private file path is exposed.