Customer safety matrix
Clear, verifiable commitments on how CloudGuard handles your Google Cloud projects and Google Workspace data. No raw credentials returned, zero email text stored, and read-only by default.
0 Email bodies stored
RFC 822 headers evaluated in RAM (<50ms) and discarded
0 Passwords recorded
Drills record input telemetry only; characters discarded
AES-256-GCM encryption
Authenticated ciphertext at rest with per-tenant derivation
Audit Manifest
Zero-Storage Invariants
Customer safety matrix8 Core Guarantees
Click "Explain to me why?" on any guarantee to launch the step-by-step visual explanation.
| Security area | Common customer concern | CloudGuard architectural protection | Safety status | Visual proof |
|---|---|---|---|---|
| Google Cloud credentials | Can CloudGuard expose our Google Cloud service account keys or OAuth tokens? | No. Keys and tokens are encrypted before storage. Your team and our app screens only see non-secret metadata, such as the service account email and connection status. | Encrypted | |
| Gmail and employee mail | Can CloudGuard store or leak employee email bodies, threads, or attachments? | No. CloudGuard scores messages in memory for phishing signals and keeps metadata only: sender, subject, message ID, time, and risk outcome. | Not stored | |
| Security awareness drills | Can CloudGuard capture real passwords employees type during phishing simulations? | No. Drills record that input happened and how many characters were typed. The password string is discarded and never sent or stored. | Not stored | |
| Tenant and client isolation | Can another customer or an unassigned employee see our findings, projects, or integration secrets? | No. Every record is tied to your organization ID. Staff only see client workspaces they are assigned to; everyone else gets an access denied response with no data. | Isolated | |
| Google Cloud changes | Can CloudGuard change or break our Google Cloud projects without approval? | No. New workspaces start in read-only audit mode. Destructive or mutating actions are blocked until an admin explicitly enables read/write remediation. | Read-only default | |
| Jira and Notion scanning | If we connect Jira or Notion, can CloudGuard expose API keys found in tickets? | No. CloudGuard stores a one-way fingerprint of a matched secret and a link to the ticket. The raw key string is discarded immediately. | Not stored | |
| Dark web breach checks | Can breach lookups send our full employee email list to third-party vendors? | No. For supported providers CloudGuard sends only a truncated SHA-256 hash prefix of each normalized email, not the address itself. | Not stored | |
| Google sign-in and delegation | Can signing into CloudGuard grant excessive Google Workspace or Cloud permissions? | No. Standard login asks only for openid, email, and profile. Workspace auditing uses separate domain-wide delegation that a super admin configures in Google Admin. | Your control |
Data boundaries at a glance
How data flows between your systems and CloudGuard. Your Google tenant stays the sovereign source of truth.
Google Cloud projects, IAM bindings, employee directory, and Gmail accounts live entirely under your sovereign administrative control.
Phishing signals and token activities are evaluated in transient memory buffers (<50ms) and immediately garbage collected.
PostgreSQL stores AES-256-GCM encrypted tokens, incident timestamps, threat labels, and deduplication fingerprints only.
Email bodies, personal messages, attachments, drill passwords, and raw third-party secrets never touch the database.
How your security team can verify or cut access at any time
You do not need to trust CloudGuard blindly. Google gives your administrators sovereign control:
1. Inspect service accounts
Open Google Cloud Console > IAM & Admin > Service Accounts. Confirm the service account and its roles match what you approved.
2. Review authorized scopes
Open admin.google.com > Security > API Controls > Domain-wide delegation. Confirm the client ID and approved OAuth scopes.
3. Revoke in one click
Delete the key in Google Cloud or remove the client ID in Google Admin. Access ceases immediately without contacting CloudGuard.
Test CloudGuard with 50 simulated enterprise identities
Verify our zero data retention, client warning isolation, and visual topology mindmap in our preloaded public demo environment. No signup or permissions required.