Candor Trust & Safety · Screening infrastructure

Kids-only should mean kids only.
Now you can prove it.

Guardian Shield Inverted screens accounts in youth spaces for adult infiltration, grooming and bullying signals — in English and Spanish — and commits every flagged case to a tamper-evident evidence ledger. Deployment logs, not promises.

See pricing Book a demo
BILINGUAL EN / ES6-LAYER COMPOSITE DETECTIONENCRYPTED EVIDENCE VAULTFAIL-TOWARD-PROTECTION
The product

Detect. Preserve. Surface.

GSI is screening infrastructure for platforms that host minors. It finds the adults who shouldn't be there — and hands you the receipts.

Detect6 signal layers

Composite scoring, cost-gated — never safety-gated.

Linguistic, behavioral, network-graph, profile, minority-verification and community-report layers score every account. A cheap bilingual pre-filter runs first; the AI model is only called when a signal trips.

PreserveAES-256 · hash-chained

Every flagged case, sealed in your own vault.

Encrypted at rest, hash-chained so any tampering breaks the chain visibly, redactions leave a tombstone, and every record is independently verifiable through the API. If evidence can't be preserved, screening fails loud.

Surfaceadult-in-the-loop

GSI detects and surfaces — it never "treats".

Each screening returns a clear verdict with reasons and an evidence receipt, routed to your parent or adult-in-the-loop workflow. Enforcement stays yours, configured per customer.

How it works

One call in. Verdict + receipt out.

Integrate in an afternoon: one screening endpoint, one verification endpoint, your API key. The ledger shows a real screening, end to end.

Integration4 steps
01

Screen

Send an account + message to POST /v1/screen. Per-message or batch, English and Spanish alike.

02

Cascade

The local bilingual pre-filter checks grooming, bullying and infiltration patterns. Only trips escalate to the AI model — standard or premium tier.

03

Verdict

A weighted composite score maps to CLEAR, MONITOR, SUSPICIOUS or ADULT_DETECTED — reasons attached.

04

Verify

Flagged cases return an evidence receipt. Anyone with the ID can confirm integrity via GET /v1/evidence/:id/verify.

The ledger — one screening, end to end
18:22:13screen requested · did demo:kAr3… · lang ES
18:22:13pre-filter trip — grooming pattern (ES): "no le digas a nadie…"
18:22:14model screening (standard tier) · adult_probability 0.94
18:22:14verdict ADULT_DETECTED (score 0.87) — routed to adult-in-the-loop
18:22:14evidence sealed · AES-256-GCM · committed at seq 41 · chain length 42
18:22:15independent verification: VALID
EVIDENCE VERIFY: VALID
{ "kind": "evidence-receipt", "vaultId": 41, "entryHash": "be68c29b6843062e0d1dfdfabded97a7…", "recordDigest": "e9230cc816810e0c0879c7c0fbda38b1…" }
Pricing

Simple plans. Metered honesty.

Every plan includes the full engine — bilingual detection, evidence vault, verification API. One screening = one account screen. Your API key and SDK are issued the moment payment completes.

Micro2,000 / mo
$49 / mo
2,000 screenings / mo included
then $0.03 per screening
Subscribe
Starter6,000 / mo
$99 / mo
6,000 screenings / mo included
then $0.025 per screening
Subscribe
Teammost popular
$149 / mo
12,000 screenings / mo included
then $0.02 per screening
Subscribe
Startup50,000 / mo
$500 / mo
50,000 screenings / mo included
then $0.015 per screening
Subscribe
Growth200,000 / mo
$1,500 / mo
200,000 screenings / mo included
then $0.01 per screening
Subscribe
Enterprisesales-led
Contact us
Custom / volume
Dedicated evidence infrastructure · custom terms · SLA
Contact sales

Overage is metered per screening beyond the included amount and billed monthly through Stripe. A higher-accuracy premium model tier is available as an add-on, billed per premium-processed message. After checkout you'll be redirected to your onboarding portal to retrieve your API key and SDK.