Complete implementation kit · KIT-EDUCATION-001

Protect learning without turning education into cognitive surveillance.

Protect inquiry by separating classroom assessment from independent research, minimizing search and learning records, prohibiting unsupported behavioral and emotional inference, preserving confidential help-seeking, requiring accessible alternatives, and reserving urgent intervention for concrete evidence of immediate harm.

Direct implementation answer

Protect inquiry by separating classroom assessment from independent research, minimizing search and learning records, prohibiting unsupported behavioral and emotional inference, preserving confidential help-seeking, requiring accessible alternatives, and reserving urgent intervention for concrete evidence of immediate harm.

People and roles

Do not collapse different people into one surveillance category.

The same technology can create different risks depending on power, age, role, location, legal authority, and the consequences of disclosure.

KIT-EDU-POP-01

Minors

Use age-appropriate support and safeguarding while preserving confidential access to health, abuse, identity, religion, politics, and other lawful information when disclosure itself may create harm.

KIT-EDU-POP-02

Adult students

Do not import parental-control assumptions into adult learning. Adult students retain privacy, due process, academic freedom, and control over ordinary intellectual exploration.

KIT-EDU-POP-03

Faculty, staff, and researchers

Protect scholarship, teaching, peer review, union activity, and professional disagreement from productivity scoring, ideological profiling, and query-based suspicion.

KIT-EDU-POP-04

Libraries and voluntary inquiry

Treat catalog searches, reference questions, database use, and voluntary reading as especially sensitive because they reveal unfinished thought rather than settled belief.

KIT-EDU-POP-05

Disability and student-support services

Do not infer disability, emotion, attention, or capacity from device behavior. Collect only information needed to provide the requested support and prevent unrelated reuse.

KIT-EDU-POP-06

Campus and school security

Respond to specific evidence of an immediate threat while prohibiting the conversion of controversial reading, protected speech, identity, or academic research into generalized danger scores.

Context boundaries

Controls must stay inside the context that justifies them.

KIT-EDU-CTX-01

Classroom teaching and assessment

Time-bounded assessment controls require notice, proportionality, accessibility, and alternatives; they do not justify continuous monitoring outside the assessment.

KIT-EDU-CTX-02

Independent scholarship and research

Researchers must be able to study disturbing, unpopular, or technically sensitive material without topic-based suspicion, while remaining accountable for unlawful conduct.

KIT-EDU-CTX-03

Confidential help-seeking

Counseling, abuse, health, disability, and crisis pathways require strict access separation and disclosure only under a documented, applicable duty tied to concrete risk.

KIT-EDU-CTX-04

Institutional administration and infrastructure

Administrative convenience, fraud control, network security, and records management must not silently become longitudinal profiles of intellectual life.

Bounded threat model

Name systems, protected activity, and credible failure paths.

Scope: Learning-management systems, student-information systems, classroom analytics, remote proctoring, AI and plagiarism detection, single sign-on, managed devices, network filtering, library systems, research platforms, disability and counseling services, automated discipline, campus security, and vendor subprocessors.

Protected activity: Lawful reading, questioning, teaching, research, drafting, political and religious exploration, confidential help-seeking, disability accommodation, association, disagreement, and use of privacy-preserving tools.

KIT-EDU-TH-01Worldwide principle

Persistent learning and inquiry dossiers

Searches, prompts, reading, drafts, library use, discussion posts, location, device activity, and assessment data can be joined into a durable profile of a learner or scholar.

KIT-EDU-TH-02Technical recommendation

Unsupported inference of attention, emotion, or dishonesty

Proctoring, analytics, and AI-detection systems may convert gaze, movement, writing style, disability, language, or network conditions into high-stakes judgments they cannot reliably support.

KIT-EDU-TH-03Worldwide principle

Topic-based threat classification

Controversial reading or research may be mistaken for endorsement, preparation, or pathology when context and concrete evidence are absent.

KIT-EDU-TH-04Worldwide principle

Confidential help-seeking exposure

Parental, administrative, vendor, or identity-provider visibility can deter minors and adults from seeking information about abuse, health, sexuality, religion, disability, or crisis support.

KIT-EDU-TH-05Operational practice

Vendor secondary use and model training

Education records may be reused for advertising, profiling, product development, model training, or unrelated risk scoring beyond the purpose students and institutions understood.

KIT-EDU-TH-06Worldwide principle

Automated discipline without meaningful review

A model-generated flag can become a disciplinary fact when reasons, evidence, context, accessibility, and human reversal are missing.

KIT-EDU-TH-07Technical recommendation

Filtering that becomes ideological or inaccessible

Broad filters can block health, identity, political, religious, disability, historical, or research material and may be hardest to challenge for the people most affected.

KIT-EDU-TH-08CognitiveLiberties.com policy proposal

Administrative and security mission creep

Identity, access, safety, and network controls introduced for a narrow event can expand into ordinary monitoring of association, belief, and inquiry.

Minimum safeguards

Build a rights-preserving floor before adding complexity.

These safeguards state the purpose that must survive local implementation. They are not a claim that every jurisdiction uses identical law or procedure.

KIT-EDU-SG-01Worldwide principle

Minimize collection at the source

Collect no query, prompt, reading event, biometric signal, derived trait, or device telemetry unless a named educational or security function demonstrably requires it.

KIT-EDU-SG-02Technical recommendation

Separate identity from inquiry

Do not require a persistent legal identity where a course role, one-time eligibility proof, anonymous credential, or local session can provide the needed access.

KIT-EDU-SG-03CognitiveLiberties.com policy proposal

Preserve a private route to knowledge

Provide a practical, accessible way to use libraries, research databases, support information, and general AI tools without creating person-level histories of lawful inquiry.

KIT-EDU-SG-04CognitiveLiberties.com policy proposal

Ban unsupported cognitive and emotional inference

Do not infer attention, emotion, disability, honesty, dangerousness, ideology, or mental condition from gaze, typing, voice, face, queries, or ordinary behavior without validated authority and a narrowly defined high-stakes need.

KIT-EDU-SG-05Worldwide principle

Require human review before adverse action

No automated flag, AI-detection score, filter event, proctoring anomaly, or threat score may independently produce discipline, exclusion, referral, or loss of opportunity.

KIT-EDU-SG-06Operational practice

Make alternatives genuinely equivalent

Offer accessible, non-punitive alternatives to invasive proctoring, biometric verification, continuous monitoring, and mandatory AI systems.

KIT-EDU-SG-07Operational practice

Bind vendors and subprocessors

Contracts must prohibit advertising, unrelated profiling, undisclosed training, onward sale, and unapproved subprocessors while requiring deletion, export, incident notice, audit, and remedy.

KIT-EDU-SG-08Technical recommendation

Use narrow, reviewable filtering

Filtering must target a defined harm, reveal reason categories, support prompt unblock review, preserve research exceptions, and be tested for language, disability, and viewpoint effects.

KIT-EDU-SG-09Operational practice

Protect confidential support contexts

Counseling, disability, abuse, health, and crisis systems must use strict role separation, short retention, documented escalation thresholds, and correction of unsupported labels.

KIT-EDU-SG-10Worldwide principle

Govern emergency intervention

Immediate intervention requires concrete evidence of a specific and credible threat or victim, a responsible human authority, the minimum necessary disclosure, and a later review.

KIT-EDU-SG-11Worldwide principle

Protect academic freedom and dissent

Policies must preserve teaching, criticism, research, satire, controversial hypotheses, and association without political loyalty tests or institutional retaliation.

KIT-EDU-SG-12Technical recommendation

Delete raw and derived records

Deletion must cover source data, prompts, recordings, embeddings, labels, scores, exported reports, backups, support copies, and downstream disciplinary records when authority ends.

Staged implementation

Move from visibility to enforceable controls to durable resilience.

First 30 daysKIT-EDU-STAGE-30

Expose data flows and stop unsupported inference

Outcome: The institution knows which systems observe intellectual activity, freezes the highest-risk unsupported uses, and publishes immediate privacy and appeal boundaries.

  1. KIT-EDU-ACT-30-01Operational practice

    Name accountable owners

    Assign named owners across education, research, libraries, accessibility, privacy, procurement, safeguarding, security, and student or faculty governance.

  2. KIT-EDU-ACT-30-02Technical recommendation

    Inventory systems and subprocessors

    Map every learning, identity, proctoring, analytics, filtering, AI, library, counseling, disability, security, and research system from collection through deletion.

  3. KIT-EDU-ACT-30-03CognitiveLiberties.com policy proposal

    Freeze unsupported adverse inferences

    Suspend automatic discipline or escalation based on emotion, attention, AI authorship, topic, disability proxy, or opaque risk scoring until evidence and due process are established.

  4. KIT-EDU-ACT-30-04Operational practice

    Publish immediate routes for privacy and appeal

    Tell people what is collected, how filters and flags work at a high level, where to seek an alternative, and who can reverse a decision.

Within 90 daysKIT-EDU-STAGE-90

Make safeguards contractual and reviewable

Outcome: High-risk systems have necessity findings, deletion schedules, accessible alternatives, vendor restrictions, and human correction procedures.

  1. KIT-EDU-ACT-90-01Operational practice

    Approve a cognitive-data classification

    Classify queries, prompts, reading, drafts, counseling, disability, biometric, and inferred-trait data as specially sensitive with role-based access.

  2. KIT-EDU-ACT-90-02Operational practice

    Rewrite vendor terms

    Amend contracts for purpose limitation, no advertising, no undisclosed training, subprocessor control, audit, deletion, export, and damages or termination for misuse.

  3. KIT-EDU-ACT-90-03CognitiveLiberties.com policy proposal

    Create independent appeal panels

    Include educational, technical, accessibility, privacy, and student or faculty representation, with authority to restore access and correct records.

  4. KIT-EDU-ACT-90-04Operational practice

    Exercise an incident scenario

    Test a false proctoring flag, exposed counseling record, overblocked research site, or compromised identity provider without using real student inquiry histories.

Within 365 daysKIT-EDU-STAGE-365

Institutionalize plural, private, and resilient learning

Outcome: Privacy engineering, academic freedom, accessible alternatives, evidence review, archive resilience, and vendor exit are recurring institutional capabilities.

  1. KIT-EDU-ACT-365-01Technical recommendation

    Build privacy into architecture

    Use local processing, short-lived identifiers, split trust, private aggregation, and inquiry-free service metrics where feasible.

  2. KIT-EDU-ACT-365-02Operational practice

    Create recurring public accountability

    Publish aggregate information on systems, retention, appeals, filter errors, incidents, vendor changes, and corrective actions without publishing intellectual histories.

  3. KIT-EDU-ACT-365-03CognitiveLiberties.com policy proposal

    Fund independent research and governance

    Support faculty, student, library, disability, and public-interest review of automated systems and protect reviewers from retaliation.

  4. KIT-EDU-ACT-365-04Technical recommendation

    Prove continuity and exit

    Demonstrate export, vendor migration, archive recovery, authentication continuity, and access to learning materials after a provider failure or termination.

Evidence and procurement checklist

Do not buy a promise. Require inspectable evidence.

A policy statement is not proof of system behavior. Require architecture, configuration, tests, contracts, logs with bounded retention, deletion evidence, and a remedy when the provider is wrong.

KIT-EDU-PROC-01Operational practice

Complete field and inference inventory

Require every raw field, event, biometric, derived inference, score, embedding, support copy, and backup to be named.

EvidenceCL-15-01
KIT-EDU-PROC-02Operational practice

Purpose and necessity evidence

Require a named educational, accessibility, security, or safeguarding purpose and explain why a less intrusive design will not work.

KIT-EDU-PROC-03Technical recommendation

Retention and verifiable deletion

Require timelines and test evidence for active data, logs, caches, models, embeddings, backups, exports, and subprocessors.

EvidenceCL-10-01
KIT-EDU-PROC-04CognitiveLiberties.com policy proposal

No advertising, sale, or unrelated profiling

Contractually prohibit behavioral advertising, sale, data-broker disclosure, and eligibility scoring from education activity.

KIT-EDU-PROC-05Operational practice

Training and product-improvement boundary

Require explicit disclosure and institution-controlled authorization before prompts, drafts, recordings, or records enter training, evaluation, or product-improvement datasets.

KIT-EDU-PROC-06Operational practice

Subprocessor and transfer register

Require advance notice, location, purpose, safeguards, deletion behavior, and objection or exit rights for every subprocessor.

EvidenceCL-10-01
KIT-EDU-PROC-07Operational practice

Model and policy change control

Require notice before changes to models, filters, thresholds, training use, retention, or adverse-decision behavior.

EvidenceCL-07-04
KIT-EDU-PROC-08Technical recommendation

Accuracy and contextual validity

Require evidence that proctoring, AI-detection, filtering, and risk tools work in the actual population and context, including language, disability, and connectivity differences.

KIT-EDU-PROC-09Worldwide principle

Human review and explainability

Require access to the evidence, broad reason categories, qualified human review, and the ability to overturn the system.

EvidenceCL-07-04
KIT-EDU-PROC-10Operational practice

Accessible equivalent alternatives

Require alternatives that do not penalize disability, poverty, shared housing, limited bandwidth, language, or refusal of invasive monitoring.

KIT-EDU-PROC-11Technical recommendation

Security and incident notice

Require independent testing, notification deadlines, containment duties, evidence preservation limits, and support for affected people.

EvidenceCL-10-04
KIT-EDU-PROC-12Technical recommendation

Export and interoperability

Require usable export of learning content, records, configuration, appeals, and accessibility data without exporting unnecessary inquiry histories.

EvidenceCL-15-01
KIT-EDU-PROC-13Operational practice

Independent audit rights

Require documentary, configuration, deletion, accessibility, security, and outcome evidence rather than relying solely on vendor attestations.

EvidenceCL-15-01
KIT-EDU-PROC-14Operational practice

Contractual remedy and exit

Require correction, deletion, service continuity, cost limits, assistance, damages where lawful, and secure destruction after termination.

Common failure modes

Good intentions can still create cognitive surveillance.

KIT-EDU-FAIL-01Technical recommendation

Treating all observation as educational evidence

More telemetry does not make an inference valid; it can instead magnify bias and make speculative labels look objective.

EvidenceCL-15-01
KIT-EDU-FAIL-02Worldwide principle

Equating surveillance with safeguarding

Universal monitoring can deter the confidential help-seeking that protects students and can expose people to the very authority causing harm.

EvidenceCL-13-02
KIT-EDU-FAIL-03Operational practice

Treating an AI score as a fact

Authorship, cheating, attention, emotion, or risk scores require context and independent evidence; they are not self-proving.

KIT-EDU-FAIL-04CognitiveLiberties.com policy proposal

Using one rule for minors and adults

A child-safety rationale does not justify parental or institutional visibility into every adult student’s inquiry.

EvidenceCL-13-04
KIT-EDU-FAIL-05Operational practice

Making invasive tools the only accessible path

A nominal opt-out is meaningless when it delays grades, costs money, exposes disability, or removes course access.

KIT-EDU-FAIL-06Technical recommendation

Keeping derived labels after deleting raw data

A deleted recording provides little protection if embeddings, risk labels, reports, and disciplinary consequences persist.

EvidenceCL-10-01
KIT-EDU-FAIL-07CognitiveLiberties.com policy proposal

Letting vendors define institutional values

Default settings and proprietary thresholds can silently decide academic freedom, discipline, privacy, and access.

KIT-EDU-FAIL-08Worldwide principle

Measuring privacy with student dossiers

A programme fails when proving compliance requires reconstructing the same person-level reading and query history it claims to protect.

Incident-response procedure

Contain concrete harm without multiplying exposure.

  1. KIT-EDU-IR-01Operational practice

    Stabilize immediate harm

    Protect a specific person or system first, using the least disclosure compatible with the immediate situation.

    EvidenceCL-02-04
  2. KIT-EDU-IR-02Worldwide principle

    Classify evidence, not the topic

    Determine whether there is a credible threat, exploitation, system compromise, policy error, model error, or protected inquiry; do not infer intent from subject matter alone.

    EvidenceCL-09-04
  3. KIT-EDU-IR-03Technical recommendation

    Stop unnecessary propagation

    Suspend automated sharing, downstream discipline, vendor synchronization, and further inference while facts are reviewed.

    EvidenceCL-07-04
  4. KIT-EDU-IR-04Operational practice

    Preserve only the minimum record

    Retain the evidence required for safety, appeal, and lawful process; avoid copying unrelated searches, drafts, counseling content, or associations.

    EvidenceCL-10-01
  5. KIT-EDU-IR-05Worldwide principle

    Notify affected people safely

    Explain the event, data, authority, temporary measures, and appeal route unless a documented legal or immediate-safety reason requires a limited delay.

    EvidenceCL-07-04
  6. KIT-EDU-IR-06Operational practice

    Require independent human review

    Use qualified reviewers outside the original automated or administrative decision chain, including accessibility or safeguarding expertise when relevant.

    EvidenceCL-38-04
  7. KIT-EDU-IR-07CognitiveLiberties.com policy proposal

    Correct and restore

    Delete unsupported inferences, correct all downstream records, restore access or standing, and address academic, financial, accessibility, or reputational harm.

    EvidenceCL-07-04
  8. KIT-EDU-IR-08Operational practice

    Publish aggregate lessons

    Record system causes, response time, corrective action, and recurrence prevention without exposing lawful inquiry or confidential support use.

    EvidenceCL-15-01
Review cadence

Make safeguards operational rather than ceremonial.

KIT-EDU-REV-01Operational practice

Monthly deletion and exception review

Review failed deletion jobs, expired legal holds, unsupported exceptions, and vendor deletion tickets.

EvidenceCL-10-01
KIT-EDU-REV-02Technical recommendation

Quarterly filter and false-positive testing

Use synthetic and consented test cases across languages, disability contexts, controversial subjects, and research use.

KIT-EDU-REV-03Operational practice

Term-by-term governance review

Review new systems, course uses, assessment tools, high-stakes flags, appeals, and accessible alternatives with learner and faculty representation.

EvidenceCL-38-01
KIT-EDU-REV-04Operational practice

Annual vendor and architecture review

Reassess necessity, subprocessors, training use, retention, model changes, security, accessibility, interoperability, and exit.

EvidenceCL-15-01
KIT-EDU-REV-05Worldwide principle

Immediate review after material change

Reopen approval after a breach, model or policy change, new inference, new population, new jurisdiction, or evidence of discriminatory impact.

EvidenceCL-02-04
Appeals and remedy

A safeguard is incomplete when no one can reverse a mistake.

KIT-EDU-REM-01Worldwide principle

Visible human appeal route

Every block, adverse flag, automated suspicion, or loss of access must point to a person or body with reversal authority.

EvidenceCL-07-04
KIT-EDU-REM-02Operational practice

Understandable reasons and evidence

Provide the policy, relevant facts, system role, uncertainty, and material relied upon without exposing protected third parties.

EvidenceCL-07-04
KIT-EDU-REM-03CognitiveLiberties.com policy proposal

Pause avoidable harm

Where immediate danger is absent, pause discipline, grade penalties, account termination, or reporting until review is complete.

EvidenceCL-09-04
KIT-EDU-REM-04Operational practice

Context and accessibility evidence

Allow the affected person to explain research purpose, disability, language, housing, connectivity, culture, or system error.

EvidenceCL-38-03
KIT-EDU-REM-05Worldwide principle

Independent review and deadlines

Set deadlines proportionate to the consequence and ensure the reviewer is independent of the original automated decision.

EvidenceCL-07-04
KIT-EDU-REM-06Technical recommendation

Downstream correction

Propagate correction to learning records, vendors, security files, faculty reports, disciplinary systems, and derived models or embeddings.

EvidenceCL-10-01
KIT-EDU-REM-07CognitiveLiberties.com policy proposal

Meaningful restoration

Restore grades, access, standing, opportunity, records, fees, and reputation where the institution caused an unsupported adverse outcome.

EvidenceCL-38-04
Data-deletion expectations

Delete the cognitive trail when the authorized need ends.

KIT-EDU-DEL-01CognitiveLiberties.com policy proposal

Search, prompt, and reading records

Do not retain identifiable lawful inquiries beyond the session or explicit user save unless a documented, narrow obligation applies.

KIT-EDU-DEL-02Technical recommendation

Learning and classroom telemetry

Delete granular clicks, dwell time, location, attention proxies, and feature events as soon as the named course function ends.

EvidenceCL-10-01
KIT-EDU-DEL-03CognitiveLiberties.com policy proposal

Proctoring media and biometrics

Use the shortest possible retention and delete video, audio, room scans, face data, keystroke data, and derived flags after review and appeal windows end.

EvidenceCL-38-03
KIT-EDU-DEL-04Technical recommendation

AI and authorship detection artifacts

Delete prompts, drafts, comparison corpora, embeddings, scores, model explanations, and unsupported labels after the bounded review purpose ends.

KIT-EDU-DEL-05Operational practice

Counseling, disability, and support data

Keep strict purpose separation and delete service metadata, derived traits, and temporary notes according to the narrowest applicable clinical, accessibility, and legal schedule.

EvidenceCL-13-02
KIT-EDU-DEL-06Worldwide principle

Security and threat-assessment records

Delete unsubstantiated tips, topic-only flags, resolved false positives, and unrelated intellectual activity; retain only what a documented safety or legal need requires.

EvidenceCL-09-04
KIT-EDU-DEL-07Technical recommendation

Backups, exports, and support copies

Ensure expiry propagates to backups, exports, support tickets, data lakes, analytics systems, and subprocessors.

EvidenceCL-10-01
KIT-EDU-DEL-08Jurisdiction-specific legal question

Legal holds and exceptions

Record the authority, scope, custodian, review date, and end condition; a hold must not suspend deletion for unrelated people or data.

EvidenceCL-02-04
KIT-EDU-DEL-09CognitiveLiberties.com policy proposal

Derived outcomes and disciplinary records

When a flag is unsupported, remove the conclusion and its downstream effects rather than merely deleting the original input.

EvidenceCL-07-04
Measurable outcomes without dossiers

Measure systems, controls, response, and recovery—not what named people think.

No metric in this kit requires an identity-linked history of lawful questions, reading, research, beliefs, associations, or use of privacy tools.

KIT-EDU-OUT-01Operational practice

System-inventory completeness

Percentage of relevant systems with named purpose, owner, fields, inferences, recipients, retention, deletion, appeal, and exit. Target: 100%.

Measurement boundary: Measure systems and contracts, not learner behavior.

EvidenceCL-15-01
KIT-EDU-OUT-02Technical recommendation

Deletion assurance

Percentage of scheduled deletion jobs and subprocessor deletion attestations completed and independently sampled on time.

Measurement boundary: Use test records and aggregate job evidence, not histories of real inquiry.

EvidenceCL-10-01
KIT-EDU-OUT-03CognitiveLiberties.com policy proposal

Private-access availability

Percentage of core library, research, support, and general learning functions offering an identity-minimized and accessible route.

Measurement boundary: Measure route availability and synthetic transactions, not who uses the route.

KIT-EDU-OUT-04Worldwide principle

Human-review coverage

Percentage of consequential automated flags receiving qualified human review before adverse action. Target: 100%.

Measurement boundary: Count decisions and process completion without retaining unrelated inquiry content.

EvidenceCL-07-04
KIT-EDU-OUT-05Operational practice

Appeal timing and restoration

Median time to acknowledge, decide, correct downstream records, and restore access or standing.

Measurement boundary: Report aggregate timing and outcome categories without publishing identities or topics.

EvidenceCL-07-04
KIT-EDU-OUT-06Operational practice

Less-intrusive assessment readiness

Percentage of courses using invasive monitoring that offer tested, accessible, equivalent alternatives.

Measurement boundary: Measure course and tool availability rather than student refusal histories.

EvidenceCL-38-03
KIT-EDU-OUT-07Technical recommendation

Synthetic filter-error rate

False blocks and missed defined harms measured with documented synthetic and authorized test sets.

Measurement boundary: Do not create a permanent log of real students’ controversial searches.

EvidenceCL-38-04
KIT-EDU-OUT-08Operational practice

Vendor accountability coverage

Percentage of relevant contracts covering training, profiling, retention, deletion, subprocessors, incident notice, audit, remedy, and exit.

Measurement boundary: Measure contract clauses and audit evidence, not vendor access to learner dossiers.

EvidenceCL-10-01
KIT-EDU-OUT-09Operational practice

Incident-response readiness

Time and task completion in exercises involving a false flag, data exposure, overblock, or vendor misuse.

Measurement boundary: Use synthetic scenarios or consented test data.

EvidenceCL-10-04
KIT-EDU-OUT-10Technical recommendation

Knowledge and archive recovery

Recovery-point and recovery-time performance for course material, research, library collections, and authenticated archives.

Measurement boundary: Measure integrity and restoration, not the identities of readers or researchers.

Jurisdiction-specific legal review

Ask these questions locally before claiming compliance.

The worldwide baseline is a rights and architecture framework. Binding duties vary across constitutions, human-rights systems, privacy, consumer, education, press, labor, accessibility, cybersecurity, records, procurement, contracts, and court procedure.

Evidence basis and limits

Research, standards, law, technical guidance, and site proposals remain distinguishable.

The kit translates supplied research, human-rights principles, academic-freedom evidence, library privacy guidance, and site policy into operational questions. It does not establish local legal compliance, validate a vendor, diagnose a student, or make a query sufficient evidence of intent.

Current law, vendor behavior, system configuration, and local risk must be independently rechecked before deployment. The exact 64-report archive remains preserved separately from this implementation derivative.

Questions institutions ask

What this kit does—and does not—require.

Does this kit prevent schools from responding to real threats?

No. It permits rapid, narrow intervention when concrete evidence indicates an immediate physical threat, active exploitation, a specific victim, or a genuine system compromise. It rejects topic-only suspicion and permanent generalized monitoring.

Does the same privacy rule apply identically to a young child and an adult researcher?

No. Safeguarding, consent, parental authority, mature-adolescent autonomy, academic freedom, employment, and research contexts differ. The kit requires those contexts to be separated rather than using one universal surveillance rule.

Can an AI-detection or proctoring score prove misconduct?

No. A score may trigger a bounded review but cannot substitute for evidence, context, accessibility analysis, a qualified human decision, and appeal.

How can an institution measure safety without retaining inquiry histories?

Measure system inventories, deletion jobs, appeal time, accessible alternatives, synthetic filter tests, vendor clauses, incident exercises, and archive recovery—not what named people read, search, ask, or believe.