If your platform allows user uploads, you need moderation you can trust—without sending sensitive content to third-party clouds. Braintrust's self-hosted moderation software runs inside your infrastructure and screens images and text before content goes live.
Why self-host content moderation
- Data stays on your servers. No images or text leave your infrastructure—critical for GDPR, internal policies, or regulated industries.
- No per-request pricing surprises. A one-time license replaces unpredictable API fees that grow with traffic.
- Lower latency. Local inference avoids round-trips to external APIs—important for real-time moderation flows.
- Full policy control. You define what is acceptable. Thresholds, categories, and escalation rules are yours to configure.
What the software moderates
Image moderation
- Detects explicit / adult visual content (nudity, graphic violence).
- Assigns a confidence score per image; you set the threshold for auto-block, review-queue, or pass.
- Processes JPEG, PNG, WebP, and GIF (first-frame analysis).
Text moderation
- Flags hate speech, harassment, profanity, and targeted abuse.
- Supports multi-language detection (Romanian, English, and other EU languages).
- Regex and keyword overrides for industry-specific terms.
AI flags; humans decide. The software generates scores and recommendations. Final publish/reject decisions remain with your team or your own automation rules—no black-box removals.
Deployment model and integration
- Runs on Linux. Docker image or direct install on Ubuntu/Debian. GPU optional (NVIDIA CUDA) for higher throughput; CPU-only mode available.
- REST API. Send an image or text payload, receive a structured JSON response with category scores. Integrate with any backend language.
- Webhook support. Get notified asynchronously when moderation completes—ideal for large-batch or queued uploads.
- Dashboard (optional). A lightweight web UI for reviewing flagged items, adjusting thresholds, and viewing analytics.
BrainTrust provides deployment guidance and can assist with integration into your existing stack.
GDPR and data protection approach
- All processing happens inside your infrastructure—no data is sent to external services.
- No telemetry, no external calls for inference. The software runs fully offline after installation.
- Logging and audit trails are configurable to meet data retention requirements.
- You control where models, logs, and queued content are stored.
Note: BrainTrust does not provide legal compliance certification. The architecture is designed to support GDPR-aligned workflows, but compliance depends on your overall data handling practices.
Licensing, updates, and support
- One-time license fee. No recurring per-request or per-user charges.
- 12 months of updates included: model improvements, security patches, and compatibility fixes.
- Optional annual maintenance (after year 1) covers continued updates and technical support—typically around 20% of the license value.
- Three tiers designed for different scales (see Pricing below).
Who it's for
- Marketplaces and classifieds — screen product photos and listing descriptions before they go live.
- Social and community platforms — moderate user-generated posts, comments, and profile images.
- SaaS products with UGC — add a moderation layer to file uploads, chat, or review features.
- Enterprise intranets and internal tools — enforce acceptable-use policies on uploaded documents and images.
- Regulated industries (healthcare, fintech, education) — keep sensitive content on-premises to satisfy compliance requirements.
Licensing & Pricing
This product is delivered as a self-hosted software license, not as a monthly SaaS subscription.
- You install it on your own infrastructure.
- You control the data.
- You avoid unpredictable per-request costs.
🟢 Launch License
Designed for startups and early-stage marketplaces.
- ✔ Self-hosted deployment package
- ✔ AI image moderation (explicit content detection)
- ✔ AI text filtering (offensive / abusive language)
- ✔ Configurable thresholds and policies
- ✔ Deployment guidance
- ✔ 12 months of updates included
Starting from €2,900 (one-time license)
🔵 Scale License (Recommended)
For growing SaaS platforms and high-volume marketplaces.
- ✔ Everything included in Launch
- ✔ Advanced policy configuration
- ✔ Performance optimization guidance
- ✔ Priority deployment support
- ✔ 12 months of updates included
Starting from €6,500 (one-time license)
🟣 Enterprise / Regulated License
For compliance-sensitive or large-scale environments.
- ✔ Custom deployment architecture
- ✔ Security hardening guidance
- ✔ Audit-ready configuration support
- ✔ Extended update & support options
Custom pricing
Annual Maintenance (Optional After Year 1)
After the first 12 months, you may continue with an optional maintenance plan that includes:
- Ongoing updates
- Security patches
- Compatibility improvements
- Technical support
Typical annual maintenance ranges around 20% of the license value, depending on tier and scope.
What We Offer
- AI image moderation — detects explicit/adult visual content with configurable confidence thresholds
- AI text filtering — flags hate speech, harassment, profanity and targeted abuse
- Multi-language support — Romanian, English and other EU languages
- REST API + webhook integration with any backend stack
- Self-hosted deployment on Linux (Docker or direct install)
- Optional GPU acceleration (NVIDIA CUDA) for high throughput
- Lightweight dashboard for reviewing flagged items and analytics
- GDPR-friendly architecture — all processing stays on your infrastructure
Benefits for Your Business
Frequently Asked Questions
What hardware do I need to run the moderation software?
The software runs on any modern Linux server. CPU-only mode works for moderate volumes. For high-throughput scenarios (thousands of images per hour), an NVIDIA GPU with CUDA support significantly improves performance. We provide sizing guidance based on your expected volume during the deployment planning phase.
How accurate is the AI detection?
The models achieve high accuracy across standard benchmarks for explicit image detection and abusive language filtering. However, no AI system is 100% accurate. That's why the software is designed around an "AI flags, humans decide" workflow — it provides confidence scores and recommendations, while final decisions remain with your moderation team or automation rules.
Can I customize what categories of content are moderated?
Yes. You can configure which categories are active (e.g., nudity, violence, hate speech, profanity), set individual confidence thresholds for each, and add custom keyword/regex rules for industry-specific terms. The system is designed to adapt to your specific moderation policy.
Is the software truly GDPR-compliant?
The architecture is designed to support GDPR-aligned workflows: all data processing happens locally, there are no external API calls or telemetry, and audit logs are configurable. However, full GDPR compliance depends on your organization's overall data handling practices. We recommend consulting your legal team for formal compliance assessment.
What happens after the 12-month update period?
After the first year, the software continues to work indefinitely — you own the license. If you want to receive continued model updates, security patches, and technical support, you can opt into an annual maintenance plan (typically around 20% of the original license value). This is entirely optional.
How does the pricing compare to cloud moderation APIs?
Cloud APIs (AWS Rekognition, Google Cloud Vision, Azure Content Moderator) typically charge $1–$4 per 1,000 images. At scale — say 500,000 images/month — that's $500–$2,000/month recurring. Our one-time license pays for itself within a few months for most platforms with significant UGC volume, with no ongoing per-request costs.