Features Overview
Comprehensive data anonymization with enterprise-grade capabilities
27001
340+ Entity Types
Comprehensive detection of personally identifiable information
Personal
- Full Names
- First Names
- Last Names
- Nicknames
- Date of Birth
- Age
- Gender
- Nationality
Contact
- Email Addresses
- Phone Numbers
- Fax Numbers
- Street Addresses
- Cities
- Postal Codes
- Countries
- GPS Coordinates
Financial
- Credit Card Numbers
- IBAN
- BIC/SWIFT
- Bank Account Numbers
- Tax IDs
- VAT Numbers
- Currency Amounts
- Crypto Wallets
Identification
- Social Security Numbers
- Passport Numbers
- Driver's License
- National ID
- Health Insurance ID
- Student ID
- Employee ID
- License Plates
Technical
- IP Addresses (IPv4/IPv6)
- MAC Addresses
- URLs
- Domain Names
- User Agents
- File Paths
- Hostnames
- API Keys
Medical
- Medical Record Numbers
- Diagnoses
- Medications
- Medical Procedures
- Lab Results
- Insurance IDs
- Hospital Names
- Doctor Names
Organizational
- Company Names
- Organization Names
- Job Titles
- Department Names
- Brand Names
- Product Names
- Project Names
- Event Names
Other
- Dates & Times
- Vehicle VINs
- Serial Numbers
- Usernames
- Passwords
- Social Media Handles
- Contract Numbers
- Invoice Numbers
5 Anonymization Methods
Choose the right protection level for your use case
Replace
Substitute PII with realistic fake data. Maintains document readability while protecting privacy.
// Before
"John Smith"
// After
"Max Mueller"Redact
Completely remove PII from the document. Irreversible but maximum privacy protection.
// Before
"john@example.com"
// After
"[REDACTED]"Hash (SHA-256)
One-way cryptographic hash. Enables consistent pseudonymization across documents.
// Before
"555-0123"
// After
"a7f3c9..."Encrypt (AES-256-GCM)
Military-grade encryption. Fully reversible with the correct key.
// Before
"DE89370400..."
// After
"enc_xK9m..."Mask
Partial visibility with masked characters. Shows structure while hiding sensitive parts.
// Before
"4532-1234-5678-9012"
// After
"4532-****-****-9012"48 Languages Supported
Including right-to-left languages for global coverage
Advanced NLP Engines
Powered by state-of-the-art machine learning models
spaCy
Industrial-strength NLP for fast, accurate entity recognition in production environments.
Stanza
Stanford's neural NLP library with state-of-the-art accuracy for 66 languages.
XLM-RoBERTa
Cross-lingual transformer model for multilingual named entity recognition.
Security & Compliance
Enterprise-grade security by design
ISO 27001:2022
Certified information security management system.
High Availability
ISO 27001:2022 certified infrastructure with monitoring and automated backups.
German Servers
Hosted on Hetzner servers in Germany. GDPR compliant.
Zero-Knowledge Auth
Argon2id password hashing. We never store your password.
AES-256-GCM
Military-grade encryption for sensitive data.
BIP39 Recovery
24-word recovery phrase for account access.
Privacy-as-Code
Structural protection — not a policy document
Traditional approach
- AI usage policy document employees ignore
- Manual checklist before every AI interaction
- DLP alerts fire after data has left
- No enforcement at the AI prompt layer
- GDPR liability if one employee slips
Privacy-as-Code with anonymize.dev
- PII interception is structural — MCP layer, not willpower
- Every AI prompt anonymised automatically
- Protection happens before data leaves your environment
- Enforced regardless of which tool the developer uses
- GDPR Art. 25 Privacy by Design — technically implemented
GDPR Art. 25 requires that data protection be built into system design. Relying on employees to manually avoid pasting PII into AI tools is not a technical safeguard — it is a policy. anonymize.dev makes PII protection a technical control: it happens automatically, at the infrastructure layer, every time.
OWASP Compliance Coverage
Which OWASP LLM Top 10 and Agentic AI risks anonymize.dev mitigates
| OWASP Risk | Category | How anonymize.dev helps | Coverage |
|---|---|---|---|
| LLM02 | Sensitive Information Disclosure | PII stripped from prompts before LLM processes them — real data never in LLM context | ● Full |
| LLM06 | Excessive Agency | Anonymised prompts reduce the value of data an over-privileged agent can access or exfiltrate | ● Partial |
| LLM07 | System Prompt Leakage | PII in system prompts anonymised via the MCP operators parameter before injection | ● Partial |
| AA3 (Agentic) | Sensitive Info Disclosure — Agentic | MCP-layer anonymisation prevents PII from entering agent context windows in multi-tool pipelines | ● Full |
| AA4 (Agentic) | Tool Misuse / Tool Poisoning | Even if a downstream MCP tool is compromised, it receives only tokens — not real PII | ● Full |
| AA7 (Agentic) | Data Exfiltration via Agent | Autonomous agents with file/DB access operate on tokenised data — exfiltration yields no real PII | ● Full |
Based on OWASP Top 10 for LLM Applications and OWASP Top 10 for Agentic AI Applications (2026 editions)
HIPAA Safe Harbor — 18 Identifiers
All 18 HIPAA Safe Harbor PHI identifiers detected and protected
HIPAA Safe Harbor method (45 CFR §164.514(b)) — all 18 identifiers removed or transformed before data reaches any AI model.
Ready to protect your data?
Start with our free tier. Access all features with any plan.