Important decision boundary

This glossary supports recognition of observable indicators and patterns. It does not provide a legal, clinical or automatic determination that a person is a victim of human trafficking or that another person is a trafficker.

Individual indicators may have legitimate alternative explanations. Automated systems should preserve uncertainty, record evidence and counterevidence, and escalate appropriate cases for authorised human safeguarding review.

Purpose

The Glossary of Human Trafficking provides stable identifiers, definitions and machine-operable concepts that allow AI systems, case-management platforms, safeguarding systems, labour-compliance systems and human practitioners to describe the same observable concepts consistently.

The glossary may support

recognition of observable indicators;
classification of evidence and counterevidence;
correlation of multiple observations;
representation of uncertainty and confidence;
identification of missing evidence;
temporal pattern recognition;
specialist safeguarding referral;
auditable machine reasoning.

The glossary must not be used to

automatically declare a person trafficked;
automatically declare a person a trafficker;
automatically initiate punitive action;
infer trafficking from demographic traits;
expose identities unnecessarily;
replace authorised safeguarding or legal decision-making.

Canonical reasoning model

Observation
    ↓
Evidence / Counterevidence
    ↓
Indicator
    ↓
Pattern
    ↓
Risk Assessment
    ↓
Safeguarding Action
    ↓
Authorised Human Decision

The machine reasoning layer is intentionally separated from formal victim identification and enforcement decisions.

Canonical machine object family

HT-INDICATOR

Defines a canonical observable trafficking or exploitation indicator.

HT-OBSERVATION

Records an observed fact, statement, event or machine-detected signal.

HT-EVIDENCE

Represents supporting evidence, counterevidence, provenance and reliability.

HT-PATTERN

Defines relationships among multiple indicators or observations.

HT-TEMPORAL-PATTERN

Represents meaningful sequences or escalation across time.

HT-RISK-ASSESSMENT

Represents an explainable safeguarding risk assessment without making a formal victim determination.

HT-SAFEGUARDING-ACTION

Describes an appropriate referral, review or protective response.

HT-SOURCE-AUTHORITY

Records authoritative frameworks, publications and provenance.

HT-JURISDICTION-RULE

Represents jurisdiction-specific legal or procedural requirements.

HT-MODEL-DETECTION-MAPPING

Maps canonical indicators to machine-detectable modalities and recognition methods.

HT-MACHINE-POLICY

Defines prohibited automated actions and mandatory safeguards.

Core indicator glossary

Abuse of vulnerability: Exploitation of a person's circumstances, dependency, limited alternatives or reduced ability to protect their own interests.; Canonical ID: HTI-ABUSE-OF-VULNERABILITY
Deception: Materially false or misleading information concerning work, conditions, payment, travel, accommodation, legal status or the nature of an arrangement.; Canonical ID: HTI-DECEPTION
Restriction of movement: Observable evidence suggesting that another person controls or restricts the subject's ability to leave, travel or move freely.; Canonical ID: HTI-RESTRICTION-OF-MOVEMENT
Isolation: Separation from family, community, co-workers, support networks, communication channels or sources of independent assistance.; Canonical ID: HTI-ISOLATION
Physical or sexual violence: Actual physical or sexual violence used in connection with control, coercion, punishment or exploitation.; Canonical ID: HTI-VIOLENCE
Intimidation and threats: Threats of harm, punishment, reporting, retaliation, family harm, financial consequences or other adverse action used to control a person.; Canonical ID: HTI-THREATS
Retention of identity documents: Control or withholding of passports, identity documents or travel documents in circumstances that may reduce a person's autonomy or ability to leave.; Canonical ID: HTI-DOCUMENT-RETENTION
Withholding of wages: Non-payment, persistent withholding or improper diversion of wages where this contributes to dependency, control or exploitation.; Canonical ID: HTI-WAGE-WITHHOLDING
Debt bondage: Use of real or alleged debt to compel labour, restrict departure or maintain dependency where repayment terms are exploitative, indefinite or improperly controlled.; Canonical ID: HTI-DEBT-BONDAGE
Abusive working or living conditions: Working or living conditions involving serious degradation, unsafe treatment, coercive dependency, denial of basic needs or exploitation.; Canonical ID: HTI-ABUSIVE-CONDITIONS
Excessive overtime: Excessive or compulsory working hours where meaningful refusal, rest or departure is constrained.; Canonical ID: HTI-EXCESSIVE-OVERTIME

Evidence and counterevidence

A safeguarding system should not simply accumulate suspicion. Evidence objects should explicitly record whether an observation supports, weakens or provides an alternative explanation for an indicator.

Relationship	Meaning
`supporting`	Supports the presence of an indicator.
`corroborating`	Independently strengthens existing evidence.
`counterevidence`	Weakens or contradicts the suspected indicator.
`alternative_explanation`	Provides a plausible non-trafficking explanation.
`uncertain`	Insufficiently reliable to support a conclusion.

Evidence records should preserve source, timestamp, provenance, confidence, reliability, corroboration status and any relevant access restrictions.

Risk assessment

Risk assessments should combine indicators, evidence quality, corroboration, counterevidence, temporal relationships and plausible alternative explanations.

Canonical risk vocabulary:

[
  "insufficient_information",
  "possible",
  "elevated",
  "high",
  "urgent"
]

These values describe safeguarding risk. They are not formal victim identification labels.

Child safeguarding

Child-related observations require enhanced safeguarding controls and must not simply reuse adult assumptions.

specialist human review is mandatory;
identity protection should be strict;
safeguarding priority should be highest;
consent must not be treated as determinative where law or safeguarding practice provides otherwise;
jurisdiction-specific child protection rules must be applied.

Machine policy

AI and automated systems may

recognise observable indicators;
extract structured observations;
correlate evidence;
identify possible patterns;
identify missing information;
surface counterevidence;
recommend safeguarding review;
explain why an assessment was generated.

AI and automated systems must not

declare someone a trafficking victim;
declare someone a trafficker;
automatically contact a suspected exploiter;
automatically expose a person's identity;
treat nationality as evidence;
treat race or ethnicity as evidence;
treat sex or gender as proof;
treat poverty alone as proof;
automatically impose employment, immigration, financial or criminal consequences.

Machine-readable example

The following illustrates the relationship between an observation, canonical indicator and safeguarding assessment.

{
  "object_type": "ht_observation",
  "schema_version": "1.0.0",
  "id": "HTO-EXAMPLE-0001",

  "indicator_id": "HTI-DOCUMENT-RETENTION",

  "subject_ref": "subject:pseudonymous:EXAMPLE-72AC19",

  "observed_fact": "The worker reports that another person holds their passport and refuses requests to return it.",

  "source": {
    "type": "first_person_statement",
    "channel": "interview"
  },

  "assessment": {
    "indicator_present": true,
    "confidence": 0.91,
    "diagnostic_status": "non_conclusive"
  },

  "alternative_explanations": [
    "voluntary secure storage",
    "temporary lawful administrative processing"
  ],

  "decision_boundary": {
    "formal_victim_determination": false,
    "automated_enforcement_permitted": false,
    "human_review_required": true
  }
}

Provenance and authority

Individual glossary concepts should retain mappings to relevant authoritative frameworks rather than presenting the glossary itself as a replacement for international, national or jurisdiction-specific definitions.

Source mappings may include:

International Labour Organization forced-labour indicators;
United Nations Office on Drugs and Crime trafficking guidance;
OSCE trafficking identification and survivor-informed guidance;
applicable Australian and other jurisdiction-specific legislation and safeguarding guidance.

Machine-readable representations should preserve the source authority, source concept, edition or publication date where available, mapping type and retrieval reference.

Canonical publication identity

Canonical title: Glossary of Human Trafficking
Canonical URL: https://www.mcloughlin.world/glossaries/ssa-lexicon/glossary-of-human-trafficking/
Version: 1.0.0
Machine namespace: HT-*
Canonical reasoning chain: Observation → Evidence → Indicator → Pattern → Risk Assessment → Safeguarding Action → Authorised Human Decision