Try an example
How It Works
Paste Your Prompt
Enter the text you’d normally send to ChatGPT, Claude, or any AI model.
Instant DLP Scan
Our engine runs 50+ detection patterns across 28+ entity categories in milliseconds.
See What’s Leaking
Get a risk report showing exactly what sensitive data types were found and their severity.
The Full AI Governance Platform
This free scanner is just a taste of what OpenSourceAIHub does for every AI request in production. Here’s what you get with the full platform:
AI Firewall & DLP Inspection
Every request passes through our real-time AI Firewall. We detect and redact 28+ entity types — emails, SSNs, credit cards, API keys, prompt injections, and more — before any data reaches the AI provider. Supports both text and OCR-based image scanning.
Smart Router & Cost Optimization
Our intelligent routing engine analyzes real-time pricing and latency across 8+ providers (OpenAI, Groq, Mistral, Together, DeepInfra, xAI, Anthropic, Google Gemini) to find the optimal path for every request. Same model, lower cost, lower latency.
Real-Time Cost Tracking
See exactly what every AI call costs — per request, per project, per day. Set budget caps that automatically block requests when spending thresholds are hit. No more surprise bills from runaway AI usage.
Flexible DLP Policy Engine
Customize protection per project: choose which entities to scan, set block vs. redact behavior, add custom regex patterns for proprietary data formats, and configure confidence thresholds. Maximum protection by default, full flexibility when you need it.
Multi-Project Architecture
Manage multiple AI applications from one account. Each project gets its own API key, DLP policy, budget limits, and usage dashboard. Ideal for agencies, teams, and enterprises running multiple AI workloads.
Wallet Mode — Managed Credits
Don't have your own AI provider keys? No problem. Add credits to your Hub wallet and we handle everything — provider selection, API calls, billing. Pay only for what you use with transparent per-token pricing.
BYOK — Bring Your Own Key
Already have API keys with OpenAI, Anthropic, or other providers? Store them securely in the Hub (AES-256 encrypted, AWS Secrets Manager backed) and we route requests through your own accounts. Zero markup. Full control.
Zero Data Retention
We never store your prompts or responses. Our stateless architecture processes and forwards — nothing is saved, nothing is logged, nothing is trained on. We only record metadata like "1 email was redacted" for your audit dashboard.
28+
Entity Types Detected
50+
Detection Patterns
<50ms
Scan Latency
8+
AI Providers Supported
Ready to Protect Every AI Request?
Deploy enterprise-grade DLP, smart routing, and cost governance in minutes. Change two lines of code — your base URL and API key — and every AI call is protected.
Frequently Asked Questions
What types of sensitive data can this tool detect?
Our scanner detects 28+ entity types including email addresses, Social Security numbers, credit card numbers, API keys (AWS, GitHub, Slack), private keys, phone numbers, physical addresses, passport numbers, driver licenses, crypto wallet addresses, IP addresses, and prompt injection attempts.
Is my text stored or logged?
No. Your text is processed in-memory and immediately discarded. We never persist, log, or train on the content you scan. Only aggregate metadata (like entity type counts) is used for the response.
How does this differ from the full OpenSourceAIHub product?
This free tool is a preview of our AI Firewall. The full platform scans every AI request in real-time, automatically redacts sensitive data before it reaches AI providers, provides per-project DLP policies, cost tracking, smart routing across 8+ providers, and a complete audit trail — all via a drop-in OpenAI-compatible API.
Why should I care about data leaking to AI providers?
When you send a prompt to an AI model, that data travels to the provider's servers. If your prompt contains PII (names, emails, SSNs), secrets (API keys, passwords), or confidential business data, that information is now outside your control. Some providers may use it for training. Regulations like GDPR, HIPAA, and SOC2 require you to prevent this.
How accurate is the detection?
Our engine combines NLP models with 50+ custom regex patterns and contextual scoring. We use a calibrated confidence threshold to minimize false positives while catching real sensitive data. The full platform allows you to tune this threshold per project.
Is OpenSourceAIHub compatible with my existing code?
Yes. If your app uses the OpenAI SDK (Python, Node.js, or any HTTP client), you only need to change two lines: the API key and the base URL. All your prompts, models, and response handling stay exactly the same.