This document outlines the technical, organizational, and procedural controls Adle AI implements to safeguard customer accounts, campaign data, and any information our AI models process.
Adle AI is a certified partner of Meta, TikTok, and Google. We operate a multi-tenant SaaS platform that creates and optimizes ad campaigns via AI agents. Security, privacy, and data isolation are foundational—every advertiser's data is segregated and never shared across accounts or competitors.
| Area | Primary Owner | Oversight |
|---|---|---|
| Security Program | Interim CISO | CEO & Board |
| Privacy & Compliance | DPO | Legal Counsel |
| Infrastructure Security | DevOps Lead | Engineering VP |
| Incident Response | Security Team | CEO |
| Layer | Method | Details |
|---|---|---|
| In transit | TLS 1.3 only | HSTS + Perfect Forward Secrecy |
| At rest | AES-256 | Managed KMS (AWS KMS) |
| Secrets | Hashicorp Vault | ADR/rotation every 90 days |
Sensitive payloads (e.g., PII, access tokens) are additionally encrypted end-to-end when travelling between micro-services using mTLS.
| Framework | Status | Notes |
|---|---|---|
| GDPR | Compliant | DPA & SCCs available |
| CCPA / CPRA | Compliant | Opt-out & deletion APIs |
| SOC 2 Type II | In progress | Target Q1 2026 · External auditor: Strike Graph |
| ISO 27001 | Planned | Gap analysis complete |
We maintain a live Sub-processor List and execute DPAs with each vendor.
A: We use OAuth with the minimal scope, store tokens in an encrypted vault, isolate each advertiser in its own micro-service agent, and rotate credentials automatically.
A: Only users you provision. Adle staff require documented, time-bound approval to access a tenant and all access is audited.
A: No. Models are fine-tuned per-tenant or on anonymised aggregates. Prompt-level guards prevent brand specifics from leaving your context.
A: Yes—TOTP or WebAuthn. Admins can enforce it organisation-wide.
A: Yes—TLS 1.3 in transit, AES-256 at rest, and optional end-to-end encryption for sensitive payloads.
Security Team: [email protected]
We commit to acknowledging any security report within 24 hours.