Home / Documentation / Security

Security

Infere's security features fall into three areas you manage yourself: account security (via AWS Cognito), API token security (IP allowlists and rate limits), and prompt security & compliance (LLM scanning and PII handling enforced at the edge).

  • Two-Factor Authentication - TOTP-based 2FA with any authenticator app, plus trusted-device management and configurable session timeouts.
  • LLM security scanning - per-token, opt-in scanning of request content for prompt injection and unsafe content against content-safety (S1–S14) and prompt-security (P1–P6) taxonomies, with a threat threshold and block-or-monitor mode.
  • PII detection & handling - per-token modes: Allow, Redact ([REDACTED]), or Block, covering emails, SSNs, card numbers, phones, and IBANs, with optional model-based detection for unstructured PII.
  • Compliance routing - restrict tokens to GDPR/HIPAA-compliant providers and enforce data residency by region (US, EU, UK, CA, AU, JP, SG, BR).
  • Edge enforcement - IP allowlists, rate limits, budgets, and token status are enforced at the edge on every request, before traffic reaches a model provider.