MediumSource Attribution◐Demonstrated
RAG source-attribution gaps
Transparency & ExplainabilityDescription
Generated answers don't reliably cite the retrieved sources, so claims can't be traced or verified.
Example scenario
A policy answer can't be traced to a source document, so staff can't verify it.
Real-world evidence◐Demonstrated
Dark patterns broadly have been confirmed in production (FTC enforcement actions) and academic research documents AI-optimised persuasion techniques, but regulatory actions specifically attributing consumer harm to AI-driven dark patterns as distinct from conventional UX dark patterns remain in early stages.
Primary mitigations
- Enforced inline citations
- cite-only-retrieved policy
- answer-source linking
- attribution checks.
Detection signals
Citation-coverage & accuracy metrics; untraceable-claim detection.
Mitigating controls
7 Non-agentic controls
ZYC-XAI-001Non-Agentic
Field Justification
Explainability (XAI)
ZYC-TRANS-001Non-Agentic
Documentation
Transparency
ZYC-ACCT-001Non-Agentic
Human-in-the-Loop
Accountability & Oversight
ZYC-XAI-005Non-Agentic
Source Citation
Explainability (XAI)
ZYC-TRANS-006Non-Agentic
Retrieval Provenance
Transparency
ZYC-TRANS-004Non-Agentic
Source Traceability
Transparency
ZYC-TRANS-005Non-Agentic
OCR Interpretability
Transparency