Add Blindfold: PII detection and protection for LLM apps

[michalvich]

What is Blindfold?

Blindfold is a PII detection and protection API built specifically for AI applications. It detects and tokenizes personally identifiable information (names, emails, phone numbers, addresses, SSNs, and more) across 18+ languages before data reaches LLMs.

Why it belongs in Frameworks for LLM security

Blindfold addresses a critical LLM security concern: preventing PII leakage to language model providers. It sits in the data pipeline before LLM calls and:

• Detects PII using NER models across 18+ languages
• Protects data via tokenization, redaction, masking, or encryption
• Enforces compliance with built-in GDPR, HIPAA, and PCI DSS policies
• Provides Python SDK (blindfold-sdk), Node.js SDK (@blindfold/sdk), LangChain integration, and MCP server
• Offers EU and US regional endpoints for data residency requirements

Placement

Added to the Frameworks for LLM security table in alphabetical order (before Plexiglass), per contribution guidelines.

• Website: https://blindfold.dev
• Docs: https://docs.blindfold.dev
• GitHub: https://github.com/blindfold-dev
• PyPI: https://pypi.org/project/blindfold-sdk/