API & Data Access

CAIM provides structured data access for government systems, research tools, and monitoring dashboards. All endpoints are publicly accessible, require no authentication, and return JSON with CORS headers enabled.

API endpoints

Endpoint	Description
/api/v3/records.json	Full record corpus — all incidents and hazards with computed fields, sources, entities, responses, and taxonomy tags
/api/v3/stats.json	Aggregate statistics — distributions by domain, severity, harm type; response coverage; hazard watch; systemic risk summary
/api/v3/systemic-analysis.json	Systemic risk analysis — risk factor co-occurrence, governance dependency gaps, cross-domain patterns, escalation metrics
/api/v3/feed.json	JSON Feed (v1.1) — chronological feed of all records for subscribers and aggregators
/api/v3/graph.json	Graph metadata — record counts, entity/system relationship maps, validation warnings, endpoint index
/api/v3/taxonomy.json	Taxonomy definitions with bilingual labels — CAIM native and OECD interoperability dimensions

All data is static JSON regenerated with each site build. The meta.generated_at field indicates when data was last built. Dates are ISO 8601 strings. Schema version: 3.1.

Records

GET /api/v3/records.json

Returns the complete record corpus. Each record includes full structured data and build-time computed fields.

Example: fetch all records

curl -s https://caim.horizonomega.org/api/v3/records.json | jq '.meta.record_count'

Incident fields

Each incident includes:

• slug, title, title_fr — identity and bilingual title
• dates — occurred (with precision), reported, occurred_end
• verification, dispute — epistemic status
• harms[] — each with bilingual description, severity, reach
• entities[] — entity slug, role primitives (developer, deployer, regulator, affected_party, reporter), bilingual description
• systems[] — system slug, bilingual involvement description
• responses[] — governance actions: actor, date, action, outcome
• sources[] — URL, title, source_type, publisher, date, language, relevance, claim_supported
• taxonomies.caim_v1 — domains, harm_types, ai_involvement, lifecycle_phases, systemic_risk_factors
• materialized_from — hazard slugs this incident traces to
• links[], aiid, version, changelog
• computed — overall_severity, reverse_links, URL

Hazard fields

Hazard objects include similar base fields plus:

• risk_pathway / risk_pathway_fr — how the hazard creates harm at current scale
• assessment_history[] — time-ordered assessments (newest first): status, confidence, potential_severity, potential_reach, evidence_summary
• escalation_model — bridging analysis: precursor_signals, governance_dependencies, catastrophic_bridge, bridge_confidence
• triggers, mitigating_factors, risk_controls
• computed — current_status, current_confidence, current_severity, current_reach, last_assessed, materialized_incidents, reverse_links, URL

Statistics

GET /api/v3/stats.json

Aggregate analytics: distributions by domain, harm type, severity, verification status, jurisdiction level, and lifecycle phase. Response coverage metrics. Active hazard watch list. Systemic risk factor summary.

Example: domain breakdown

curl -s https://caim.horizonomega.org/api/v3/stats.json | jq '.distributions.domains'

Systemic risk analysis

GET /api/v3/systemic-analysis.json

CAIM's distinctive analytical output: cross-record patterns connecting deployment-level failures to catastrophic risk trajectories.

• risk_factor_counts — frequency of each systemic risk factor across all records
• cooccurrence — pairs of factors that cluster on the same records (top 20)
• governance_gaps — institutional capacities extracted from hazard escalation models, ranked by cross-domain spread
• cross_domain — factors in 3+ domains, indicating structural properties transcending sector governance
• escalation — per-hazard metrics: assessment count, status transitions, bridge confidence
• bridge_confidence_distribution — low/medium/high across all hazards

Example: top governance gaps

curl -s https://caim.horizonomega.org/api/v3/systemic-analysis.json | jq '.governance_gaps[:3]'

Feed

GET /api/v3/feed.json

A JSON Feed v1.1 combining all incidents and hazards in reverse chronological order. Subscribe in any feed reader or monitoring tool.

Integration examples

Government procurement check

Before deploying an AI system, a department can query CAIM for relevant hazards in their domain:

# Fetch all health-domain incidents
curl -s https://caim.horizonomega.org/api/v3/records.json \
  | jq '.incidents[] | select(.taxonomies.caim_v1.domains[]? == "health") | {title, computed}'

Risk dashboard integration

Pull aggregate statistics into an internal monitoring dashboard:

# Get current breakdown by domain and severity
curl -s https://caim.horizonomega.org/api/v3/stats.json \
  | jq '{summary, distributions: {domains: .distributions.domains, severity: .distributions.severity}}'

Systemic risk monitoring

Track the most common structural failure patterns across Canadian AI deployments:

# Get top risk factor co-occurrences
curl -s https://caim.horizonomega.org/api/v3/systemic-analysis.json \
  | jq '.cooccurrence[:5] | .[] | "\(.factor_a) + \(.factor_b): \(.count) records"'

OECD-compatible export

Every record carries OECD AIM interoperability fields alongside CAIM's native taxonomy. The taxonomy endpoint (/api/v3/taxonomy.json) exposes both CAIM native and OECD dimension definitions with bilingual labels. Records are structured for contribution to the OECD's open submission pathway. See the Methodology for the two-layer taxonomy design.

Terms of use

CAIM data is published as a public-interest resource. Government agencies, researchers, journalists, and civil society organizations may use the API freely. Automated access is welcome — CAIM data is designed to be consumed programmatically.

If you build on CAIM data, please cite the monitor. See individual record pages for citation formats.