Skip to main content

Mission and Public-Interest Framing

World Transparency Graph exists to improve public-interest oversight through structured analysis of public records. The platform is designed for transparency, accountability, and responsible investigative workflows.
Policy-Version: v1.0.0
Effective-Date: 2026-02-28
Owner: WTG Governance Team

Prohibited Uses

The following uses are strictly forbidden:
  • Doxxing, intimidation, or harassment of individuals
  • Political persecution or coordinated targeting based on ideology
  • Extortion, blackmail, or coercive use of graph outputs
  • Publication of personal data outside approved public-safe policy
  • Treating analytical signals as automatic legal proof without judicial determination
Violations may lead to:
  • Access restrictions
  • Investigation lockouts
  • Internal incident review and policy escalation

Human-Review Requirement for High-Risk Allegations

Any high-risk claim must go through human editorial/legal review before publication or escalation. High-risk claims include:
  • Corruption allegations
  • Fraud allegations
  • Reputationally sensitive claims

Minimum Review Requirements

Before publishing or escalating high-risk claims:
  1. Confirm supporting evidence from independent sources
  2. Validate temporal consistency and identity confidence
  3. Add limitations and uncertainty statements

Review Checklist

  • Multiple independent sources confirm the relationship
  • Temporal ordering is verified (event before consequence)
  • Identity confidence is documented (strong vs probable)
  • Uncertainty and limitations are explicitly stated
  • Neutral, non-accusatory language is used

Non-Accusatory Language Standard

BR-ACC outputs must use neutral language that reflects the analytical nature of the platform. Use neutral, investigative language:
  • “signal”
  • “possible connection”
  • “documented relationship”
  • “requires further verification”
  • “suggests potential link”
  • “indicates association”

Forbidden Framing

Do NOT use definitive criminal conclusions without judicial determination:
  • “guilty of”
  • “proven fraud”
  • “criminal”
  • “corrupt” (without court conviction)
  • “illegal” (without legal determination)

Disclaimer: Patterns Are Signals, Not Proof

Graph patterns indicate relationships in public data. They do not by themselves prove:
  • Illegality
  • Intent
  • Guilt
  • Criminal activity
BR-ACC output is not legal advice and is not a substitute for formal legal or investigative process.

Source Quality Limitations

Public datasets may contain:
LimitationImpact
Delays and incomplete periodsMissing recent data or historical gaps
Source-level schema changesInconsistent field definitions across time
Identifier ambiguityMultiple entities with similar names
Masked fieldsPartial identifiers for privacy protection
Documentation gapsMissing metadata or context
Extraction errorsTechnical issues in ETL pipelines

False-Positive Caveat

All high-impact conclusions require independent corroboration from multiple sources.

Temporal and Identity Confidence

Interpretation must account for:

Temporal Ordering

  • Event before consequence: Verify that relationships existed before outcomes
  • Coverage windows: Understand which time periods are included
  • Missing-month caveats: Document known gaps in temporal coverage

Identity Quality Tiers

TierDescriptionUse Case
StrongUnique identifier match (CNPJ, official ID)Direct entity attribution
ProbableName + context matchRequires human verification
WeakName-only matchInvestigation starting point only
When uncertainty exists, outputs must be framed as hypotheses requiring further verification.

What BR-ACC Is Not

BR-ACC is not:
  • A criminal accusation system
  • A credit scoring platform
  • An automated guilt determination tool
  • A replacement for legal due process
  • A substitute for professional legal advice

What BR-ACC Is

BR-ACC is:
  • A research and analysis tool
  • A public data aggregator
  • An investigative starting point
  • A transparency and accountability platform
  • A civic technology project for public interest

Enforcement and Governance

Violations of this ethics policy are handled according to:

Incident Review Process

When ethical violations are suspected:
  1. Immediate review by governance team
  2. Access restriction if necessary to prevent harm
  3. Investigation into scope and intent of violation
  4. Decision log with rationale and action taken
  5. Policy escalation if systemic issues are identified

Responsible Use Principles

Always cite sources and document the origin of data connections. Never present derived insights as definitive facts without source attribution.
Access only the data necessary for your legitimate research or investigative purpose. Do not engage in bulk collection or speculative data gathering.
Take responsibility for how you use and share BR-ACC outputs. Consider potential harms and misinterpretations.
Ensure your use of the platform is proportional to the public interest served. Do not use powerful analytical tools for trivial or personal purposes.

Build docs developers (and LLMs) love