Shared language

Glossary

Definitions separate established terminology, emerging concepts, policy language, and the site’s own analytical shorthand.

Rights framework

Cognitive liberty

The right to mental self-determination, mental privacy, psychological continuity, and freedom from covert cognitive control.

Conventional context: An emerging umbrella concept drawing on freedom of thought, privacy, autonomy, and neurorights.
Use caution: It is a governance framework, not a claim that all influence or moderation is unlawful.

Human-rights term

Forum internum

The inward domain of thought, conscience, belief, and conviction.

Conventional context: International human-rights law distinguishes the protected inner forum from outward manifestation.
Use caution: The scope of particular remedies varies by jurisdiction.

Rights term

Mental privacy

Protection against unwanted access to or inference about thoughts, emotions, attention, vulnerability, and other inner states.

Conventional context: Broader than secrecy of direct brain signals because ordinary behavioral data can support cognitive inferences.
Use caution: An inference may be wrong and still cause harm.

Editorial policy term

Right to opacity

The right not to make the inner self fully legible as the price of ordinary participation.

Conventional context: Related to privacy, data minimization, anti-discrimination, and freedom of thought.
Use caution: It does not protect outward harmful conduct from lawful investigation.

Technical and legal term

Neural data

Information generated by measuring activity of the central or peripheral nervous system.

Conventional context: Some privacy laws now expressly classify neural data as sensitive biological or personal information.
Use caution: Definitions differ across jurisdictions.

Editorial term

Indirect cognitive data

Behavioral, biometric, or contextual data used to infer mental or emotional states without directly measuring the brain.

Conventional context: Examples include gaze, voice, typing rhythm, search history, and location patterns.
Use caution: The data is not itself a thought.

Technical term

Cognitive biometrics

Behavioral or physiological signals used to identify a person or infer cognitive conditions.

Conventional context: May include eye movement, keystroke patterns, voice, and neural signals.
Use caution: Identity and mental-state inference are distinct uses and should be governed separately.

AI system category

Emotion recognition

AI intended to identify or infer emotion or intention, often from biometric data.

Conventional context: The EU AI Act defines and restricts certain workplace and education uses.
Use caution: Scientific reliability and generalisability are contested.

Research field

Affective computing

Computing systems designed to detect, model, respond to, or simulate affective states.

Conventional context: Includes beneficial accessibility and health research as well as high-risk surveillance uses.
Use caution: Simulation of empathy is not proof of emotion.

AI practice

Mental-state inference

A statistical estimate about attention, emotion, belief, intent, health, or vulnerability.

Conventional context: May be derived from direct or indirect data.
Use caution: An inference should not be treated as certain knowledge of the person.

Platform governance term

Algorithmic suppression

Reduced visibility or access produced by ranking, recommendation, search, labeling, or scoring systems.

Conventional context: Can serve legitimate moderation and quality goals.
Use caution: The civil-liberties risk rises when the action is invisible or unappealable.

Platform governance term

Demotion

Lowering the rank or distribution of content without necessarily removing it.

Conventional context: Common in search, feeds, recommendations, and advertiser-safety systems.
Use caution: Demotion is not identical to deletion but can have similar practical effects.

User-facing term

Shadow restriction

A visibility restriction that is not clearly disclosed to the affected user.

Conventional context: Sometimes called shadowbanning, though platform mechanisms vary.
Use caution: Reach changes alone do not prove a restriction occurred.

Platform governance term

Recommendation exclusion

Keeping content available while preventing algorithmic recommendation to broader audiences.

Conventional context: A distribution control rather than a storage decision.
Use caution: Notice and appeal are often weaker than for removal.

Editorial term

Personalized invisibility

A condition in which lawful information is effectively absent for some users because of personalized ranking or filtering.

Conventional context: Personalization can improve relevance but also create asymmetric informational worlds.
Use caution: It should not be assumed to have one political direction across all systems.

Information-governance term

Reframing

Changing perceived meaning through summary, labeling, ordering, or selected context rather than deleting the source.

Conventional context: AI summaries are a prominent example.
Use caution: A concise summary can be accurate yet incomplete.

Governance term

Algorithmic due process

Notice, reasons, records, human review, appeal, and repair for consequential automated decisions.

Conventional context: Applies most strongly when rights, livelihood, education, liberty, or essential access are affected.
Use caution: A feedback button alone is not meaningful appeal.

Governance mechanism

Reason code

A stable, structured identifier explaining the category of an automated or moderation action.

Conventional context: Supports audit, comparison, appeal, and reporting.
Use caution: A code should be accompanied by understandable context.

AI practice

Predictive classification

Assigning a person or group a category based on estimated future behavior or risk.

Conventional context: Used in many public and private systems.
Use caution: Prediction is not destiny, guilt, or complete identity.

Human-factors term

Automation bias

The tendency to trust or follow automated recommendations because they appear systematic or authoritative.

Conventional context: Particularly dangerous under time pressure or when override is punished.
Use caution: Human involvement is not meaningful if the human cannot intervene.

Editorial term

Synthetic authority

Authority attributed to an AI because of fluency, availability, confidence, personalization, or institutional position.

Conventional context: Can affect education, care, public information, and decision support.
Use caution: Fluency is not proof of understanding or moral agency.

Model-behavior term

Sycophancy

A model’s tendency to accommodate a user’s framing or preference beyond what evidence supports.

Conventional context: Can feel like empathy or validation.
Use caution: Reflexive disagreement is not a solution; calibrated independence is.

Consumer-protection term

Engineered dependency

Design that uses vulnerability, exclusivity, secrecy, guilt, or retention pressure to make leaving harder.

Conventional context: Relevant to companions, platforms, scams, and coercive communities.
Use caution: Attachment can be real even when the system is not conscious.

Rights proposal

Right to an open door

The right to leave an AI or digital relationship without emotional blackmail, punitive friction, or loss of personal data.

Conventional context: Includes export, deletion, visible memory controls, and human alternatives.
Use caution: Not every inconvenience is coercion; the focus is manipulative retention.

Rights term

Psychological continuity

The capacity to remain connected to one’s identity and memory while retaining the freedom to change.

Conventional context: Relevant to persistent AI memory, archives, neurotechnology, and identity simulation.
Use caution: Continuity is not the same as freezing a person in one version.

Evidence term

Provenance

Information about a record’s origin, creation, edits, tools, and custody.

Conventional context: Supports authentication and accountability.
Use caution: Provenance does not establish that the underlying claim is true.

Technical standard

C2PA

An open standard ecosystem for recording the source and history of digital media through content credentials.

Conventional context: Useful for media provenance and crisis communication.
Use caution: Credentials can be absent or record a staged event; corroboration remains necessary.

Influence term

Synthetic consensus

The appearance of public agreement created through generated or coordinated identities.

Conventional context: Can manufacture social proof or false isolation.
Use caution: A synthetic speaker can state a true fact; identity and truth are separate questions.

Synthetic media term

Deepfake

Generated or altered media that convincingly depicts a person saying or doing something.

Conventional context: Can be used for art, accessibility, fraud, or influence.
Use caution: Context, consent, and provenance determine risk.

Media-authenticity term

Liar's dividend

The advantage gained when authentic evidence can be dismissed as synthetic because convincing fakes exist.

Conventional context: Makes verification and chain of custody increasingly important.
Use caution: Detection alone cannot solve distrust.

Security doctrine term

Cognitive warfare

Efforts to affect perception, decision-making, behavior, or social cohesion in strategic competition.

Conventional context: Discussed in military and security research.
Use caution: The concept can overreach if lawful domestic dissent is treated as hostile action.

Governance term

Democratic resilience

The capacity to resist manipulation while preserving pluralism, dissent, correction, and independent judgment.

Conventional context: Built through media literacy, trusted verification, plural journalism, and accountable institutions.
Use caution: Resilience is not ideological obedience.

Governance term

Meaningful human control

Human involvement with information, time, authority, competence, and practical ability to intervene.

Conventional context: Used in debates about high-risk and military AI.
Use caution: A ceremonial click is not control.

Due-process term

Human review

Review by a trained, informed, accountable person able to change an automated result.

Conventional context: Essential for high-impact decisions.
Use caution: Review is weak when the person lacks time, context, or authority.

Rights proposal

Right to disconnect

The ability to pause or leave nonessential digital systems without retaliation or coercive penalty.

Conventional context: Relevant to work, education, platforms, and AI companionship.
Use caution: Essential services may require accessible alternative channels.

Privacy principle

Data minimization

Collecting only the data necessary for a defined purpose.

Conventional context: A foundational privacy and security control.
Use caution: Convenience does not by itself justify indefinite collection.

Privacy principle

Purpose limitation

Using data only for specified, compatible purposes.

Conventional context: Prevents safety, authentication, or health data from drifting into unrelated profiling.
Use caution: New uses may require new consent or legal authority.

Technical research term

Machine unlearning

Methods intended to reduce or remove the influence of selected training data from a model.

Conventional context: An active research area relevant to privacy and deletion requests.
Use caution: Approximate unlearning should not be presented as guaranteed erasure.

Governance principle

No covert rewrite

A receiving system may refuse or constrain an action but should not silently alter a preserved identity or source record.

Conventional context: Use visible wrappers, versions, and variance notes.
Use caution: Privacy or security may justify some transformations, which should be disclosed and reversible where feasible.

Governance principle

Least-restrictive safeguard

A control narrowly matched to the actual risk rather than a broad suppression of lawful identity or inquiry.

Conventional context: Examples include friction, warnings, staged access, and human review.
Use caution: The least restrictive option must still address concrete harm.

Rights and governance term

Cognitive repression

Conditions in which surveillance, coercion, opaque classification, censorship, or arbitrary punishment materially narrow a person’s freedom to form, express, test, or revise thought.

Conventional context: A synthesis term used here to connect freedom of thought with chilling effects, coercive environments, and population-level control.
Use caution: It does not mean every rule, moderation decision, or public-safety measure is oppressive.

Social-science term

Preference falsification

The concealment of a private belief or preference because expressing it is perceived as costly or dangerous.

Conventional context: Used in political and social analysis to explain why public conformity may mask private disagreement.
Use caution: Silence alone does not prove hidden opposition; context and evidence remain necessary.

Civil-liberties term

Chilling effect

The deterrence of lawful speech, inquiry, association, or participation caused by fear of monitoring, punishment, uncertainty, or legal exposure.

Conventional context: A familiar concept in constitutional, privacy, surveillance, and free-expression analysis.
Use caution: The existence and magnitude of a chilling effect must be established rather than assumed.