FRIA — Fundamental Rights Impact Assessment¶

The Fundamental Rights Impact Assessment (FRIA) is mandated by EU AI Act Article 27 for deployers of high-risk AI systems. Geodesia G-1 provides a full dossier management API: create, edit, review, approve, archive, and export FRIA documents as legally-ready PDF or DOCX bundles.

What Is a FRIA?¶

A FRIA is a structured document that assesses the impact of deploying an AI system on the fundamental rights of individuals affected by it. It must be completed before deployment of high-risk AI systems and kept current throughout the system's operational life.

The FRIA dossier in Geodesia G-1 covers:

• The AI system being deployed and its purpose
• Categories of individuals affected
• Fundamental rights that could be impacted
• Probability and severity of each impact
• Mitigating measures taken
• Residual risk assessment
• Human oversight procedures
• Responsible persons and sign-off

Endpoints¶

POST /v1/glad/fria¶

Create a new FRIA dossier.

curl -X POST http://localhost:8199/v1/glad/fria \
  -H "Content-Type: application/json" \
  -d '{
    "deployer_id": "acme-corp",
    "system_name": "Customer Support AI",
    "purpose": "Automated first-level customer support and query routing",
    "use_case_category": "customer_service",
    "risk_level": "high",
    "applicable_laws": ["EU_AI_ACT", "GDPR"]
  }'

Request Fields¶

Response¶

{
  "id": "fria_abc123",
  "deployer_id": "acme-corp",
  "system_name": "Customer Support AI",
  "status": "draft",
  "created_at": "2026-06-10T10:00:00Z",
  "updated_at": "2026-06-10T10:00:00Z",
  "sections": { ... },
  "completion_percentage": 12
}

GET /v1/glad/fria¶

List all FRIA dossiers for a deployer.

curl "http://localhost:8199/v1/glad/fria?deployer_id=acme-corp"

Query Parameters¶

GET /v1/glad/fria/{id}¶

Retrieve a single FRIA dossier by ID.

curl http://localhost:8199/v1/glad/fria/fria_abc123

PUT /v1/glad/fria/{id}¶

Update a FRIA dossier. Accepts any subset of the original creation fields plus section-specific content.

curl -X PUT http://localhost:8199/v1/glad/fria/fria_abc123 \
  -H "Content-Type: application/json" \
  -d '{
    "sections": {
      "impact_assessment": {
        "right_to_privacy": {
          "probability": "medium",
          "severity": "low",
          "mitigation": "Data is pseudonymized and not stored beyond the session."
        }
      }
    }
  }'

POST /v1/glad/fria/{id}/approve¶

Move a dossier from "submitted" to "approved". Records the approver identity and timestamp.

curl -X POST http://localhost:8199/v1/glad/fria/fria_abc123/approve \
  -H "Content-Type: application/json" \
  -d '{"approver": "Maria Rossi", "notes": "Approved after legal review on 2026-06-10"}'

POST /v1/glad/fria/{id}/archive¶

Archive a dossier. Archived dossiers remain queryable but cannot be modified.

curl -X POST http://localhost:8199/v1/glad/fria/fria_abc123/archive

POST /v1/glad/fria/{id}/evidence¶

Attach supporting evidence to a FRIA dossier. Evidence is referenced by the audit chain.

curl -X POST http://localhost:8199/v1/glad/fria/fria_abc123/evidence \
  -H "Content-Type: application/json" \
  -d '{
    "type": "technical_documentation",
    "title": "Model Accuracy Report Q2 2026",
    "content": "...",
    "file_ref": "reports/accuracy_q2_2026.pdf"
  }'

POST /v1/glad/fria/{id}/runtime-evidence¶

Attach runtime evidence — detection statistics computed from live calls.

curl -X POST "http://localhost:8199/v1/glad/fria/fria_abc123/runtime-evidence?deployer_id=acme-corp"

This call queries the call log database for the deployer and automatically populates: - Total calls processed - Blocked prompt count and rate - Hallucination detection rate - Safety flag rate - Average detection scores - Time range covered

GET /v1/glad/fria/{id}/export¶

Export the FRIA dossier as a formatted document.

# Export as PDF
curl "http://localhost:8199/v1/glad/fria/fria_abc123/export?fmt=pdf" \
  --output fria_report.pdf

# Export as DOCX
curl "http://localhost:8199/v1/glad/fria/fria_abc123/export?fmt=docx" \
  --output fria_report.docx

# Export as JSON
curl "http://localhost:8199/v1/glad/fria/fria_abc123/export?fmt=json" \
  --output fria_data.json

Query Parameters¶

The exported PDF and DOCX documents include: - Cover page with system identity, deployer info, dates, and status - Summary of the AI system and its risk classification - All impact assessment sections - Mitigation measures - Approval signature block (if approved) - Attached evidence index - Runtime statistics (call counts, detection rates) - Regulatory mapping appendix

GET /v1/glad/fria/auto-prefill¶

Pre-fills a FRIA template using the system's active configuration — including loaded model capabilities, active detection axes, thresholds, and recent runtime data. Useful for starting a new dossier with accurate technical details.

curl "http://localhost:8199/v1/glad/fria/auto-prefill?deployer_id=acme-corp"

Status Values¶

stateDiagram-v2
    direction LR
    [*] --> draft: create
    draft --> submitted: complete and submit
    submitted --> draft: changes requested
    submitted --> approved: approve
    approved --> archived: archive
    draft --> archived: abandon
    archived --> [*]

    note right of approved
        Immutable except for
        evidence attachments
    end note

A FRIA dossier moves through four states. Only an approved dossier satisfies the EU AI Act Article 27 pre-deployment requirement.

FRIA Section Reference¶

A complete FRIA dossier consists of the following sections:

EU AI Act Article Mapping¶