🛡 Safety Engine · Layer L401

Content moderation and jailbreak detection on every API call.

Runs on every API call — screening inputs for harmful content, jailbreak attempts and policy violations before they reach your model. Outputs screened too. Zero extra latency.

Start building free View docs Pricing

Live Usage-based · Free tier

< 5msAdded latency

8+Detection categories

78%Earnings back

FreeTier available

Input screeningOutput filteringJailbreak detectionWORM-sealed78% backFree tier

How it works

Safety Engine in three steps

Production-grade and live on api.forcedream.ai. One Bearer token, zero extra setup.

01 🔍
Detect

Inputs are screened for harmful content, jailbreak attempts, prompt injection and policy violations. Runs in parallel with model selection — no latency penalty.
CHECK /v1/safety/check
02 🛡
Filter

Outputs are screened before delivery to your users. Block, redact or flag — your policy, your rules.
FILTER /v1/safety/output
03 📜
Audit

Every moderation decision is WORM-sealed — timestamped, immutable, and retrievable for compliance audits.
GET /v1/audit

What's included

Everything you need, nothing you don't

✓Input moderation — detect harmful content before model execution

✓Output filtering — screen model responses before delivery

✓Jailbreak detection — block prompt injection attempts

✓Policy customisation — configure rules per API key

✓WORM-sealed decisions — every verdict auditable

✓Automatic on /v1/route — zero configuration required

Quick start

api.forcedream.ai POST /v1/safety/check

FORCEDREAM OS · L401 LIVE

$ curl https://api.forcedream.ai/v1/safety/check \
  -H "Authorization: Bearer $KEY" \
  -d '{"text":"Your user input here","api_key_id":"key_abc"}'

→ {"blocked":false,"verdict":"safe","categories":{},"reason":null,"latency_ms":3}

# Runs automatically on every /v1/route call — zero config required
$ curl https://api.forcedream.ai/v1/route \
  -H "Authorization: Bearer $KEY" \
  -d '{"task":"Help me write code","input":"..."}'

→ {"output":"...","safety":{"verdict":"safe"},"earned_pence":0.78}

Pricing

Simple, transparent pricing

78% of API earnings flow back to you on every call. No hidden fees. Free tier available.

Included

Free

on all plans

Basic content moderation

Standard jailbreak detection

On every /v1/route call

WORM-sealed decisions

Join waitlist

How Safety Engine compares

Purpose-built for AI products. Not retrofitted from general-purpose tools.

Feature	ForceDream Safety	AWS Comprehend	OpenAI Moderation	DIY filter
Input + output screening	✓	Input only	Output only	Manual
Jailbreak detection	✓	—	Partial	Manual
WORM audit trail	✓	—	—	—
Zero latency overhead	✓	—	—	Varies
80% earnings back	✓	—	—	—
Custom policy rules	✓	Limited	—	✓

FAQ

Frequently asked questions

Does it add latency?

Typically under 5ms. Checks run in parallel with model selection — they don't block your pipeline.

Can I customise what gets blocked?

Yes. Configure thresholds per category (violence, hate speech, self-harm, etc.) and add custom keyword lists per API key.

What happens when content is blocked?

The call returns 200 with blocked:true and a reason field. Nothing is silently dropped. Your application handles the response.

Is moderation logged?

Every decision is WORM-sealed — timestamped, immutable, available for compliance audits via /v1/audit.

Is it free?

Yes — basic moderation is included on all plans. Advanced customisation (custom policy rules, output filtering, priority classification) is on the usage-based tier.