Ship child safety.
One endpoint.
The Aletheia API is currently in closed beta. General availability opens to developers soon. Request early access and we'll reach out when your seat is ready.
How it works
POST a conversation. Get back a structured assessment: harmfulness score, predatory-user identification, risk tier, confidence, and latency. That's it.
Authentication
Every request carries a bearer token. Create and revoke API keys from your account dashboard once you're approved.
-H "Authorization: Bearer sk-aletheia-..." \
-H "Content-Type: application/json"
/v2/predict
Single Prediction
Score one conversation in real time. Use this for live moderation on chat, DM, and messaging platforms.
- Sub-50ms p95 latency
- Multi-task output: harmfulness, predator ID, risk tier
- Per-request pricing on standard tier
/v2/predict/batch
Batch Processing
Queue thousands of conversations for asynchronous scoring. Ideal for historical audits, backfills, and trust & safety review queues.
- Webhook or polling delivery
- Volume-based enterprise pricing
- Conversations scored and discarded — zero retention
Two tiers. One mission.
Pricing reflects operational cost, not profit. Polycreek is a 501(c)(3) nonprofit — every request funds continued research into child safety AI.
Per-request
Pay for what you use. Ideal for startups, indie developers, and platforms testing integration.
- Single-prediction endpoint
- Community rate limits
- Email support
- Zero CSAM stored, ever
Annual subscription
For platforms, law enforcement agencies, and institutions integrating Aletheia at scale.
- Batch + streaming endpoints
- Dedicated throughput and SLA
- Optional in-house NCMEC reporting by Polycreek analysts
- Custom data-processing agreements
Join the early-access list.
Be first in line when the Aletheia API opens to the public.
Request Early Access