Security

Security at waai

Protecting your data — customer accounts, WhatsApp conversations, and connected integrations — is foundational to waai. This page describes the security controls we operate, which are aligned with the AICPA Trust Services Criteria — the Security Common Criteria (CC6.x, CC7.x) and the Availability criteria (A1.x) — each control area below notes the criteria reference it maps to. An independent SOC 2 audit is planned; until it is completed, this page describes our controls and is not a certification or audit report.

Encryption at Rest (CC6.1)

• Encrypted credentials. WhatsApp access tokens, AI provider API keys, and OAuth tokens (including Google) are encrypted at rest using Fernet symmetric (AES) encryption and are never stored in plaintext.
• Password hashing. Account passwords are hashed with bcrypt and salted — they are never stored or transmitted in readable form.

Multi-Factor Authentication (CC6.1)

• Authenticator apps. Every account can enable time-based one-time passwords (TOTP) with any authenticator app.
• Passkeys. Passwordless sign-in with passkeys (WebAuthn / FIDO2) — fingerprint, face, or hardware key.
• Backup codes. Single-use recovery codes so a lost device never locks you out.

Role-Based Access Control (CC6.2 & CC6.3)

• Granular roles. Team access is managed through Owner, Admin, Customer Service Officer, and Custom roles with per-page permissions.
• Tenant isolation. Every query is scoped to your account — one business's data is never visible to another.

Secure Integrations & Transmission (CC6.6 & CC6.7)

• HTTPS/TLS everywhere. All traffic to and from waai is encrypted in transit.
• Verified webhooks. Incoming events from Meta, Stripe, and Cal.com are validated with HMAC-SHA256 signature verification; Apple server notifications are JWS-verified with certificate-chain pinning. Forged or unsigned requests are rejected.
• Minimum-scope OAuth. Integrations such as Google Workspace are connected with the narrowest scope set the feature needs, and you can revoke access at any time.

Application Hardening (CC6.6)

• SSRF protection. Every server-side fetch of a user-supplied URL passes an outbound guard that blocks internal and private network targets.
• Rate limiting. API endpoints are rate-limited per IP and per user to blunt brute-force and abuse.
• Strict input validation. All request bodies are schema-validated; uploads are restricted by type and size.
• Prompt-injection defense. User text that reaches AI models is sanitized, model outputs are filtered, and system prompts are hardened against leakage.

Data Backup & Recovery (A1.1 & A1.2)

• Automated daily platform backups. The entire platform database is snapshotted automatically every day on a grandfather-father-son schedule — the last 7 daily, 4 weekly, and 6 monthly copies are retained — so both recent mistakes and older incidents are recoverable.
• On-site and off-site copies. Snapshots are stored with restricted file access and can be replicated to off-site cloud storage.
• Your configuration, your copy. From Settings → Backup & Restore, download a complete backup of your bot configuration (knowledge base, flows, bookings, catalogs, contacts, and settings), restore from a platform-kept snapshot, or import a bundle as new items.
• Integrity-verified restores. Every snapshot is SHA-256-verified before it is used, and backups never contain plaintext credentials — sensitive fields stay encrypted exactly as they are stored.

SOC 2 Alignment & Planned Independent Audit

Contact

Questions about security or reporting a vulnerability? Contact us at info@waai.me.