What our products see, what leaves your browser, what we store — and for how long
This statement complements (does not replace) the SafeToOpen Privacy Policy and DPA.
The short version
SafeToOpen analyzes web pages, links, downloads, and reported emails to detect zero-day phishing. We are designed around data minimization: analysis happens in the browser first, and only when a page requires VisionAI analysis do its visual elements and URL leave the device. Browsing-history checks use one-way hashes, so your actual browsing history is never exposed to us. Our models are not trained on customer data. All customer data is hosted in New Zealand. We never sell data or use it for advertising. We are ISO/IEC 27001:2022 certified and GDPR compliant.
Products covered
SafeToOpen browser extension (Chrome, Edge, Firefox), Email Verification, Customer Email Verification, Brand & Customer Protection, and the SafeToOpen API.
What the extension can access, and what actually leaves the browser
The browser stores require us to declare our permissions and data practices; our declared categories are website content, personally identifiable information, and location data, with binding commitments that data is not sold to third parties and not used for purposes unrelated to the product's core function. The table below explains why each category is needed and what actually happens with it.
| Data category | Why the product needs it | What leaves the browser | Stored by SafeToOpen? | Retention |
|---|---|---|---|---|
| Page code & structure (DOM, scripts) | X-Ray inspects page structure for hidden threats | Nothing — X-Ray analysis runs locally in the browser | No | — |
| Page visual appearance | VisionAI compares what the user sees — logos, branding, login forms — against impersonation patterns | Visual elements of the page, only when the page cannot be cleared locally and requires VisionAI analysis | Yes, pending analysis | Deleted within 30 days; confirmed-malicious pages retained as threat intelligence |
| URL of the analyzed page | Analysis context and threat lookup | The URL, only alongside a VisionAI submission | Yes, with the submission | Same as above |
| Browsing history (Security Checkup) | Detect past exposure to compromised sites | One-way hashes of URLs only — actual browsing history is never exposed to SafeToOpen | No — hashes are checked and discarded | Transient |
| Form input / paste events (Paste Guard) | Prevent credentials being entered into untrusted pages | Nothing — evaluated locally; the sensitive value itself is never transmitted | No | — |
| Downloads | Block malicious installers, malware, ransomware | Files are not uploaded; checks use the source URL and threat intelligence | No | — |
| Email address (breach checkup) | Check against known breach databases | The email address, for the check only | Not retained beyond performing the check | Transient |
| Reported emails (Email Verification products) | Analyze employee/customer-reported suspicious emails, links, and attachments | The reported message, submitted by the customer for analysis | Yes — for analysis and audit trail | 90 days, then deleted; confirmed-malicious indicators retained as threat intelligence |
| Installed extensions list (Security Checkup) | Flag known-malicious extensions | Checked against threat data | No | Transient |
| Account & licensing data (name, email, plan) | Provide and bill the service | Standard account data | Yes | Life of account + 30 days (billing records kept 7 years per NZ tax law) |
| Product telemetry / interaction data | Reliability and product improvement | Optional technical data, as declared in the browser store listing | Yes | 12 months |
What we never do: sell or rent data; use browsing data for advertising or profiling; read or store form values, passwords, or keystrokes; upload users' files.
Where processing happens
First-pass analysis (X-Ray, Paste Guard, local checks) runs in the user's browser. When a page requires VisionAI analysis, its visual elements and URL are transmitted over TLS to the SafeToOpen cloud, hosted in New Zealand data centres operated by SiteHost. Customer data submitted for analysis is processed and stored in New Zealand. A data-flow diagram is included in the SafeToOpen Security Overview.
AI/ML and your data
SafeToOpen's detection models (VisionAI and X-Ray) are trained on data we gather ourselves — publicly reachable phishing and legitimate pages collected by our own systems and confirmed threat intelligence. Customer data is not used to train or improve our models. Confirmed phishing pages we discover may be shared with the security community (e.g., blocklist operators) to protect the wider public; this concerns attacker infrastructure, not customer data. See the SafeToOpen AI & Machine Learning Transparency Statement for detail.
Enterprise deployment controls
Business customers can contact us to configure deployment to their policy needs, including exclusion of internal domains from analysis so internal applications are never submitted for cloud analysis, and alert visibility for administrators. Contact [email protected] to discuss deployment configuration for your environment.
Legal bases, rights and applicable law
SafeToOpen Ltd (Auckland, New Zealand) acts as processor for customer-submitted content under our DPA and as controller for account data under our Privacy Policy. We support data subject rights (access, correction, deletion, portability) under the NZ Privacy Act 2020, GDPR, and the Australian Privacy Principles. Data is hosted in New Zealand, which holds a European Commission adequacy decision, meaning transfers from the EU/EEA to New Zealand are lawful without additional safeguards. Requests: [email protected].
Sub-processors
The current list, locations, and purposes are published in the SafeToOpen Sub-Processor List on our Trust Center. Customers are notified at least 30 days before a new sub-processor processes customer data.
Security of the data we hold
Controls protecting stored data — encryption in transit (TLS 1.2+) and at rest (AES-256), least-privilege access with MFA, monitoring, incident response, and breach notification within 72 hours of confirmation — are described in the SafeToOpen Security Overview and certified under our ISO/IEC 27001:2022 ISMS.
Questions
Privacy: [email protected] · Security: [email protected] · General: [email protected]
Document control: v1.0 · Approved by the Managing Director · Next review: July 2027. Related: Security Overview · AI Transparency Statement · Privacy Policy · DPA · Terms of Service.