What Gateway does.
Every capability is designed to work autonomously — detecting, responding, and protecting without human intervention.
🧠
Layer 1 — Linguistic Age Analysis
Claude AI analyzes language patterns to detect writing that indicates a minor. Not keyword matching — genuine linguistic pattern recognition across vocabulary, syntax, and communication style.
📊
Layer 2 — Behavioral Patterns
Posting times, content interests, activity patterns, and session behavior. A 13-year-old's usage pattern looks measurably different from an adult's — the system learns what those patterns look like.
👤
Layer 3 — Profile Signals
Username patterns, bio content, and photo signals are evaluated. Subtle indicators that a profile belongs to a minor are detected across multiple signal dimensions.
🕸️
Layer 4 — Network Graph
Connection analysis identifies accounts linked to other flagged accounts. Minors cluster in network graphs — the system maps these patterns in real time.
🤖
Layer 5 — Proof-of-Humanity
Bot prevention ensures the account belongs to an actual person before deeper analysis. Automated account creation for exploitation is blocked at this layer.
🔐
Layer 6 — Zero-Knowledge Age Verification
Users prove they are 18+ to a trusted verifier. Their ID never touches Candor's servers. The platform stores only: age_over_18: true + a cryptographic hash of the proof.
🏳️
Layer 7 — Community Reporting
Users can report suspected minors, feeding signals back into the detection pipeline. Community intelligence supplements AI detection.