Signals, not verdicts. Evidence-linked. Human-reviewed. The index summarizes disclosed posture using public documentation and timestamps every published update.
Posture signals
Axis-level score signals based on disclosed posture.
Confidence
HIGH, MED, LOW reflect evidence coverage and clarity.
Evidence links
Published updates reference primary/public documentation.
Update timestamps
Every published change includes timing and review context.
Each pillar captures one part of disclosed human safety posture.
How clearly an entity discloses policy boundaries, enforcement pathways, and misuse controls.
How clearly data collection, retention, sharing, and user control options are disclosed.
How clearly security architecture, incident response posture, and vulnerability handling are disclosed.
How clearly impersonation protections, youth safeguards, and reporting/escalation controls are disclosed.
Signals use a 0-5 posture scale per pillar. Higher scores indicate stronger disclosed posture based on publicly available documentation. Scores are not guarantees of outcomes and may be incomplete when disclosures are limited.
Multiple recent primary sources corroborate the same posture signal.
Evidence exists but is partial, older, or not fully corroborated.
Evidence is sparse, ambiguous, or not recent enough for strong confidence.
Step 1
Sources monitored
Step 2
Snapshots captured
Step 3
Diffs generated
Step 4
Draft updates created
Step 5
Admin review and approval
Step 6
Published to changefeed
Every entity page includes a corrections form. Submissions enter the admin review queue and only affect public signals after explicit review.