Dated evidence maintenance

Research changes should be visible.

This page records which sources were reopened, what was checked, which wording decisions changed or stayed the same, what remains unresolved, and when evidence should be reviewed again. It is a maintenance record—not a claim that every report or every sentence has been independently verified.

Direct answer

Release 1.12.0 preserves the existing 31 bounded source reviews without claiming a new external verification round. It records a direct handoff of 32 reports as exact duplicates of the then-existing 48-report archive, so no second source copy or competing authority was created. The ledger records 64 reviewed sources, 20 explicitly unrechecked sources, and 0 blocked sources. Exact report, package, artwork, translation, and Advanced Persona authority bytes remain outside the new implementation derivatives. Exact custody for sixteen additional AI-governance reports is recorded separately in the 1.12.0 intake receipt, expanding the immutable archive to 64 reports without claiming that those new report assertions have already passed the established 194-unit evidence review.

Current evidence status

Source freshness is separated from source authority.

A primary court record, official technical standard, peer-reviewed study, provider document, and civil-society report have different authority even when all are recently retrieved. The counts below describe review state only.

64

Reviewed online

Source identity and the named support scope were checked on 2 September 2026.

20

Not reopened

Known locators remain in the ledger without a new online review event.

0

Blocked or unavailable

These records remain explicit gaps rather than being counted as verified.

No scorecard: V-Dem, Freedom House, Access Now, OONI, and the Academic Freedom Index use different methods, periods, and units. Their findings can support a worldwide trend analysis, but they are not merged into one country ranking.
31 source checks

What was actually reopened

Each event records the source, date, method, checked scope, and limit. The event identifier remains stable for later correction or supersession.

V-Dem Institute — Democracy Reports

REV-2026-09-02-001

Confirmed the current Democracy Reports publication surface and the 2026 report used for the worldwide freedom-of-expression trend statement.

Limit: The project confirmed source identity, current public availability, and the specific comparative trend used by the site. It did not independently reproduce V-Dem’s full statistical model or country coding.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official project publication page review · outcome: current public record confirmed · checked 2026-09-02

Freedom House — Freedom on the Net

REV-2026-09-02-002

Confirmed the Freedom on the Net 2025 report and the report-body count used by the worldwide overview: 28 of 72 assessed countries deteriorated and 17 improved.

Limit: The review confirms the source’s published comparative count and scope. It does not treat the 72-country study as a complete measure of every country or of cognitive liberty as a whole.

Conflict note: A contradictory summary snippet was not adopted; the project retained the figure stated in the report body and records the source-specific methodology boundary.

Method: authoritative report page review · outcome: current public record confirmed with count reconciliation · checked 2026-09-02

Access Now — #KeepItOn coalition

REV-2026-09-02-003

Confirmed the coalition’s 2025 shutdown findings used on the worldwide page: at least 313 shutdowns across 52 countries.

Limit: Shutdown counts depend on the coalition’s definitions, measurement access, partner reporting, and attribution methods. They do not measure all forms of censorship.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: authoritative civil society report review · outcome: current public record confirmed · checked 2026-09-02

OONI — open measurement of internet censorship

REV-2026-09-02-004

Confirmed OONI’s current project-level measurement scale used by the site: more than three billion measurements, approximately thirty thousand networks, and 245 countries and territories since 2012.

Limit: The review confirms the project’s published measurement coverage. Individual measurements require country, network, test, and anomaly-level interpretation before attributing censorship.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official measurement project review · outcome: current public record confirmed · checked 2026-09-02

Academic Freedom Index

REV-2026-09-02-005

Confirmed the 2026 update, its global comparative coverage, and the project’s five-indicator approach to academic freedom.

Limit: Composite scores and change estimates depend on expert coding, confidence intervals, edition boundaries, and the project’s published methodology.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official project methodology and update review · outcome: current public record confirmed · checked 2026-09-02

UN Human Rights Committee — General Comment No. 34 on Article 19

REV-2026-09-02-006

Confirmed General Comment No. 34 as the Human Rights Committee’s authoritative interpretive guidance on Article 19 freedoms of opinion and expression.

Limit: General Comment No. 34 is authoritative treaty-body guidance, but application to a specific legal dispute still requires jurisdictional and factual analysis.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official un human rights record review · outcome: stable authoritative record confirmed · checked 2026-09-02

UN Special Rapporteur — encryption and anonymity in digital communications

REV-2026-09-02-007

Confirmed the 2015 Special Rapporteur report treating encryption and anonymity as enabling a protected zone for opinion and expression and applying legality, necessity, and proportionality to restrictions.

Limit: The report is expert UN guidance rather than a self-executing global statute. National implementation and later jurisprudence differ.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official un document record review · outcome: stable authoritative record confirmed · checked 2026-09-02

OHCHR — Rabat Plan of Action

REV-2026-09-02-008

Confirmed the Rabat Plan of Action and its high threshold for restricting incitement, including contextual evaluation of speaker, intent, content, extent, and likelihood.

Limit: The Rabat framework guides rights-respecting analysis; it does not erase jurisdiction-specific law or decide a case without evidence about context and intent.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official ohchr outcome document review · outcome: stable authoritative record confirmed · checked 2026-09-02

UN Guiding Principles on Business and Human Rights

REV-2026-09-02-009

Confirmed the UN Guiding Principles’ protect, respect, and remedy framework and the responsibility of businesses to conduct human-rights due diligence.

Limit: The Guiding Principles are globally influential standards, not a single directly enforceable legal code in every jurisdiction.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official un guiding principles review · outcome: stable authoritative record confirmed · checked 2026-09-02

UNESCO Recommendation on the Ethics of Artificial Intelligence

REV-2026-09-02-010

Confirmed the UNESCO Recommendation on the Ethics of Artificial Intelligence as a 2021 global standard adopted by 193 Member States and centered on human rights, proportionality, privacy, transparency, accountability, and human oversight.

Limit: Adoption establishes an international normative framework; implementation, enforcement, and national legal effect vary.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official unesco standard page review · outcome: current authoritative standard confirmed · checked 2026-09-02

Internet Society — internet fragmentation resources

REV-2026-09-02-011

Confirmed Internet Society’s framework describing fragmentation as divergence that can impair the open, globally connected, secure, and trustworthy Internet.

Limit: The framework supports architectural analysis, not attribution of a specific outage or policy to censorship without separate evidence.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official technical policy resource review · outcome: current authoritative guidance confirmed · checked 2026-09-02

Open Source AI Definition 1.0

REV-2026-09-02-012

Confirmed Open Source AI Definition 1.0 and its requirements that users be able to use, study, modify, and share an AI system with access to the preferred form for modification.

Limit: The definition establishes an open-source standard; it does not by itself prove that a particular model or package conforms.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official standard definition review · outcome: current standard confirmed · checked 2026-09-02

Global Network Initiative — Principles on Freedom of Expression and Privacy

REV-2026-09-02-013

Confirmed the Global Network Initiative principles and implementation guidance on company responsibilities for freedom of expression and privacy when facing government restrictions or demands.

Limit: The principles are a governance commitment and assessment framework; participation or publication does not automatically prove complete operational compliance.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official multistakeholder principles review · outcome: current governance principles confirmed · checked 2026-09-02

Santa Clara Principles on Transparency and Accountability in Content Moderation

REV-2026-09-02-014

Confirmed the Santa Clara Principles’ procedural commitments concerning understandable rules, notice, numbers, appeals, cultural competence, and transparency about state involvement in moderation.

Limit: The principles are a civil-society accountability benchmark rather than a universal binding legal rule.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official principles record review · outcome: current governance principles confirmed · checked 2026-09-02

UNESCO — Safety of Journalists

REV-2026-09-02-015

Confirmed UNESCO’s current safety-of-journalists programme and its role in monitoring, capacity building, and implementation of the UN Plan of Action on the Safety of Journalists and the Issue of Impunity.

Limit: Programme materials support the global institutional framework; incident-level claims require separate source and jurisdiction review.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official unesco programme review · outcome: current authoritative programme confirmed · checked 2026-09-02

UNESCO — Memory of the World Programme

REV-2026-09-02-016

Confirmed the Memory of the World programme’s purpose of preserving, protecting, and improving access to documentary heritage.

Limit: The programme supports preservation principles and registers documentary heritage; it does not independently verify every archive or historical claim cited by the site.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official unesco programme review · outcome: current authoritative programme confirmed · checked 2026-09-02

Tor Project — censorship circumvention support

REV-2026-09-02-017

Confirmed the Tor Project’s current censorship-circumvention support documentation and the project’s privacy and anti-censorship purpose.

Limit: First-party documentation establishes intended functionality and current support guidance; availability and effectiveness vary by network, device, threat model, and jurisdiction.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official project support documentation review · outcome: current first party documentation confirmed · checked 2026-09-02

Signal — protocol and cryptographic specifications

REV-2026-09-02-018

Confirmed the current public Signal protocol specification index, including key-agreement and ratcheting specifications used to document end-to-end encryption architecture.

Limit: Specifications document protocol design. A particular implementation, deployment, endpoint, or operational practice requires separate verification.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official technical specification review · outcome: current technical specifications confirmed · checked 2026-09-02

CLEAR Global — language access and crisis communication

REV-2026-09-02-019

Confirmed CLEAR Global’s current public work on language access, multilingual communication, and language technology for crisis-affected and underserved communities.

Limit: The organization’s public record supports the importance of language access; project-specific effectiveness and local translation quality require separate evaluation.

Conflict note: No unresolved source-identity conflict found during this bounded review.

Method: official language access organization review · outcome: current project record confirmed · checked 2026-09-02

ACLU — the Constitution in the 100-mile border zone

REV-2026-09-02-020

Confirmed that the ACLU page was updated June 8, 2026 and describes the federal government’s claimed 100-mile border-zone authority while stating that important Fourth Amendment protections still apply.

Limit: This is current U.S. civil-liberties guidance, not primary law, legal advice, or a worldwide standard. Statutes, regulations, facts, and circuit-specific precedent require separate review.

Conflict note: No source-identity conflict found. The implementation kit does not generalize this jurisdiction-specific example into a universal legal rule.

Method: official civil liberties guidance review · outcome: current guidance confirmed with jurisdiction boundary · checked 2026-09-02

American Library Association — privacy guidelines for library management systems

REV-2026-09-02-021

Confirmed guidance on library data ownership, clear policies, opt-in use, user access and deletion, collection minimization, non-perpetual retention, encryption, restricted reports, government-request procedures, API profiling risk, vendor audits, and incident-response planning.

Limit: The guidance was last revised in April 2022 and is professional guidance rather than binding law. Local legal duties, contract language, system behavior, and vendor security require independent review.

Conflict note: No source-identity conflict found. Its recommendations are labeled as professional guidance and do not override jurisdiction-specific law.

Method: official professional guidance review · outcome: current professional guidance confirmed · checked 2026-09-02

American Library Association — minors and online activity

REV-2026-09-02-022

Confirmed the June 29, 2025 amended interpretation supporting minors’ expression and privacy, age-appropriate resources, safety education, limits on excessive filtering, noncommercial treatment of minors’ activity, and individual responses to misuse.

Limit: The interpretation is an ALA professional policy position, not a controlling legal rule in every library, school, or jurisdiction. Minors’ records and access rights require local legal analysis.

Conflict note: No source-identity conflict found. The library kit keeps youth safety, privacy, and local authority questions in separate layers.

Method: official professional policy interpretation review · outcome: current policy interpretation confirmed · checked 2026-09-02

Google Search Quality Rater Guidelines

REV-2026-09-02-023

Confirmed the official September 11, 2025 guideline edition and its Page Quality and Needs Met framework, including page purpose, creator and site responsibility, reputation, E-E-A-T, YMYL, query meaning, locale, and user intent.

Review locator: publisher or institutional record used for this bounded check.

Limit: Google states that individual rater judgments do not directly move pages and that the guidelines are not a complete ranking formula. The review does not infer undocumented ranking signals or guarantee discoverability.

Conflict note: The canonical raterhub locator resolves to Google’s current static PDF. No contradiction requiring public wording change was found.

Method: official current guideline pdf and search central review · outcome: current official guideline confirmed · checked 2026-09-02

IEEE 7000 standards family overview

REV-2026-09-02-024

Confirmed the current IEEE Global Initiative 2.0 page, its continued IEEE 7000 Series awareness work, and its stated focus on toolkits, conformity and impact assessment, generative AI, and practical ethics.

Review locator: publisher or institutional record used for this bounded check.

Limit: The programme page describes standards activity and intended resources. It does not establish compliance, certification, safety, or rights-respecting behavior by any particular deployed system.

Conflict note: The original URL now redirects to the current IEEE programme page. The source remains a standards-family overview, not a product endorsement.

Method: official ieee standards programme review · outcome: current programme and standards family confirmed · checked 2026-09-02

Ranking Digital Rights methodology

REV-2026-09-02-025

Confirmed the current RDR methods surface and archive: an open methodology with 58 human-rights indicators in governance, freedom of expression, and privacy, expanded into more than 300 questions about transparency and normative practices.

Review locator: publisher or institutional record used for this bounded check.

Limit: RDR primarily evaluates company commitments, disclosure, and governance. It does not by itself prove production behavior, hidden system operation, or the experience of every user in every jurisdiction.

Conflict note: The 2022 route is preserved as source identity while the current methods page is recorded as the review locator. The kit uses RDR as an accountability method, not operational proof.

Method: official methodology and archive review · outcome: methodology scope confirmed · checked 2026-09-02

On Neurorights

REV-2026-09-02-026

Confirmed the 2021 peer-reviewed Frontiers article, its publication identity, and its conceptual review of neurorights, including cognitive liberty and related mental protections.

Review locator: publisher or institutional record used for this bounded check.

Limit: The article is scholarly analysis in a developing and contested field. It is not binding law and does not establish universal consensus or adoption of every proposed neuroright.

Conflict note: The PubMed Central locator was inaccessible to the review client, but the publisher full text and PDF resolved the source identity and scope without changing the original source URL.

Method: publisher full text and publication metadata review · outcome: peer reviewed source identity and scope confirmed · checked 2026-09-02

Marthews and Tucker — Government Surveillance and Internet Search Behavior

REV-2026-09-02-027

Confirmed the working paper’s authorship, eleven-country Google Trends design, April 29, 2015 dated copy, and reported decline in selected privacy-sensitive searching after publicity about government surveillance.

Review locator: publisher or institutional record used for this bounded check.

Limit: The analysis is observational, uses selected terms and aggregate search data, and cannot eliminate all alternative explanations or establish an effect for every person, topic, country, or surveillance system.

Conflict note: The SSRN record resolved, while the dated full text was reviewed through an accessible copy. The site preserves association and causal caveats.

Method: ssrn record and accessible working paper review · outcome: working paper identity method and findings confirmed · checked 2026-09-02

Jon Penney — Chilling Effects: Online Surveillance and Wikipedia Use

REV-2026-09-02-028

Confirmed the 2016 article, institutional repository record, interrupted time-series design, and reported immediate decline and changed longer-term trend in views of privacy-sensitive Wikipedia articles after the surveillance disclosures.

Review locator: publisher or institutional record used for this bounded check.

Limit: The study is observational and cannot eliminate every alternative explanation. Its measured Wikipedia behavior should not be generalized mechanically to all information services or surveillance contexts.

Conflict note: The author’s institutional repository resolved the source identity and abstract scope after the original SSRN locator was inaccessible to the earlier review client.

Method: institutional repository record and article scope review · outcome: peer reviewed article identity method and findings confirmed · checked 2026-09-02

Neil Richards — Intellectual Privacy

REV-2026-09-02-029

Confirmed the article’s definition of intellectual privacy as protection for records of intellectual activities and its argument that such protection supports freedom of thought and expression.

Review locator: publisher or institutional record used for this bounded check.

Limit: The article is legal scholarship and normative analysis. It is not binding law, and application to a particular records system, subpoena, institution, or jurisdiction requires independent legal review.

Conflict note: The author’s university repository supplied the full text after the original SSRN page had been inaccessible to the earlier review client.

Method: author institutional repository full text review · outcome: legal scholarship identity and core thesis confirmed · checked 2026-09-02

The concept of function creep

REV-2026-09-02-030

Confirmed Bert-Jaap Koops’s 2021 peer-reviewed Law, Innovation and Technology article and its conceptual analysis of changes in a system’s or authority’s proper activity commonly described as function creep.

Review locator: publisher or institutional record used for this bounded check.

Limit: The concept helps identify contestable repurposing. It does not prove that every secondary use is illegitimate, intentional, or irreversible; specific systems require factual and legal analysis.

Conflict note: The publisher page was inaccessible to the earlier review client, but the author’s institutional research record resolved publication identity and scope. The existing source ID and URL remain stable.

Method: author institutional repository metadata and abstract review · outcome: peer reviewed source identity and conceptual scope confirmed · checked 2026-09-02

Hypernudging — Big Data as a mode of regulation by design

REV-2026-09-02-031

Confirmed Karen Yeung’s 2017 peer-reviewed article and its account of networked, continuously updated, dynamic, data-driven choice architecture that can channel attention and decisions.

Review locator: publisher or institutional record used for this bounded check.

Limit: The article supplies a socio-legal concept and rights-based critique. It does not prove that every recommender, personalization system, or nudge is coercive or harmful.

Conflict note: The publisher page was inaccessible to the earlier review client, but the King’s College London record and accepted-manuscript link resolved source identity and scope without renumbering the source.

Method: author institutional repository record and accepted manuscript scope review · outcome: peer reviewed source identity and conceptual scope confirmed · checked 2026-09-02

Editorial change control

Wording decisions and conflicts

Source reports remain immutable. A public correction or clarification is documented separately so readers can distinguish new evidence from editorial judgment.

REC-2026-09-02-001

Source count confirmed no public number change

Issue: A secondary search-result summary conflicted with the count in the report body.

Resolution: Use the report-body count already published by the site: 28 of 72 assessed countries deteriorated and 17 improved. Preserve the explicit boundary that this is one source’s comparative sample, not a world score.

Public effect: No numerical correction was required; evidence provenance and methodology language were strengthened.

REC-2026-09-02-002

Cross source scope clarification applied

Issue: Global trend indicators use different units, coverage periods, and methods and could be misread as one composite country score.

Resolution: Keep each figure attached to its source and state that the combined pattern is an editorial inference about worldwide pressure, not a merged index or a ranking of populations.

Public effect: The worldwide and update pages now repeat the no-scorecard and methodology-separation rule at the point of use.

REC-2026-09-02-003

Translation authority boundary published

Issue: A translated summary could be mistaken for an independently authoritative or perfectly equivalent source.

Resolution: Publish the Spanish pilot as a clearly labeled derivative, link to the controlling English page, disclose AI assistance and editorial review, list omissions, and state that independent native-speaker review remains pending.

Public effect: The Spanish page and translation manifest expose language, source revision, transformations, omissions, review state, and integrity hash.

REC-2026-09-02-004

Alternate authoritative locator review resolved previous blockers

Issue: Six scholarly source locators were previously classified as blocked or unavailable even though publisher or author-institution records could resolve their identity and bounded support scope.

Resolution: Preserve every stable SRC identifier and original URL, add the reviewed alternate publisher or institutional locator to the event ledger, and classify the source as verified only for publication identity, method, and named support scope. Keep observational, conceptual, and nonbinding-authority limitations explicit.

Public effect: The blocked queue falls to zero without upgrading scholarship into law, treating correlation as universal causation, or rewriting any archived report.

REC-2026-09-02-005

Implementation kit authority layers and non surveillance metrics published

Issue: A practical kit could accidentally collapse worldwide rights principles, professional guidance, technical standards, local legal duties, operational advice, and this site’s policy proposals into one falsely authoritative checklist. Metrics could also reward identity-linked monitoring of lawful reading.

Resolution: Publish every kit item with one explicit recommendation class; isolate jurisdiction-specific questions; require local legal review; bind evidence references to stable CL and SRC records; and define outcome boundaries that prohibit patron-level histories of lawful inquiry, reading, searches, or beliefs.

Public effect: The library kit is actionable without becoming legal advice, a country scorecard, vendor endorsement, or surveillance programme. Its public page and JSON source expose the same stable IDs and authority classes.

Deterministic maintenance queue

What needs attention next

The repository’s standard-library tool classifies every source and every evidence unit from committed metadata. It does not browse the web. The queue is therefore reproducible, reviewable, and safe to run offline.

Source and evidence-unit queues as of 2026-09-02
StatusSource recordsEvidence unitsMeaning
Overdue00Records whose next-review dates precede the current as-of date.
Due soon134Reviewed records scheduled within the next 90 days.
Blocked or unavailable00Source locators that could not be directly retrieved during the recorded review.
Not reopened2056Known source locators carried forward without an independent online recheck in this release.
Current51134Reviewed records whose next-review dates fall beyond the current review window.

Blocked source records

No source record remains in the blocked queue in release 1.12.0. Six previously inaccessible scholarly locators were resolved through publisher or author-institution records, with the original source URLs and all evidentiary limits preserved.

Inspect all 84 source records

Evidence into practice

Four complete implementation kits keep authority layers separate.

The implementation-kit programme now publishes complete kits for libraries, education, journalism and exile media, and public-interest technology. Each draws on existing section-level evidence without treating human-rights records, standards, technical research, professional guidance, and site policy as interchangeable. Every safeguard, procurement question, incident step, deletion expectation, and outcome measure is labeled as a worldwide principle, a jurisdiction-specific legal question, a technical recommendation, an operational practice, or a CognitiveLiberties.com policy proposal.

Stable derivative identity

KIT-LIBRARIES-001, KIT-EDUCATION-001, KIT-JOURNALISM-001, KIT-PUBLIC-INTEREST-TECH-001, and every component ID remain bound to their public pages and machine-readable records.

No legal overclaim

Local law, collective agreements, records duties, minors’ rights, accessibility, procurement, and disclosure authority remain questions for qualified local review.

No surveillance metric

Success measures may count aggregate completion, response time, deletion proof, appeal access, or audit coverage—but not identity-linked histories of lawful reading, searching, borrowing, or belief.

See the implementation-kit programme and future sectors

Governed multilingual access

The first Spanish pilot is a derivative, not a new authority.

The published Spanish worldwide summary is bound to source revision 1.7.0 and the controlling English page. It discloses AI assistance, editorial review, transformations, omissions, and the absence of independent native-speaker review.

Translation ID
TR-ES-001
Review state
editorially reviewed derivative native review pending
Integrity
SHA-256 exact artifact bytes
Native-speaker review
Pending
Reproducible without a network

How the freshness queue works

  1. Read committed facts. The tool loads the claim ledger, source-package receipt, and mechanism taxonomy.
  2. Verify package custody. It checks the exact received ZIP’s size and SHA-256 before generating a queue.
  3. Classify deterministically. Any blocked or unrechecked states take precedence; reviewed sources are then classified by next-review date.
  4. Reconcile every identifier. All 84 sources and all 194 evidence units must appear exactly once.
  5. Write no factual conclusion. A queue entry says what to review, not whether the underlying claim is true.