Data security and retention
Where reports are kept, how they are protected, who can see them and how long we keep them.
Where reports are kept
Reports and attachments are stored in a private storage area in Australia (Sydney), provided by our hosting provider, Vercel. The website’s server functions also run in Sydney. Reports are never stored in our content system, which holds public pages only.
How we protect them
- Encryption: information is encrypted in transit (HTTPS) and at rest by our storage provider.
- Hidden details removed: your browser redraws photos and screenshots before upload, and our server redraws them again when they arrive, removing location, device details and the original file name.
- Tamper evidence: we keep a digital fingerprint (SHA-256) of each attachment, so we can show it has not been altered since you sent it.
- File checks: only image files are accepted, checked by their content and not just their name.
- Limits: we limit how often reports can be sent, to stop automated misuse.
- Private storage: attachments cannot be opened by a public link. Only signed-in caseworkers can view them.
- No report data on devices: the app does not keep reports on your phone after you send them.
- New report alerts: caseworkers receive an alert with the reference number and type of report only, never its contents.
Who can access reports
Only caseworkers trained in privacy, trauma-informed practice, child safety and our classification method can read reports. They have regular supervision and debriefing, because reading hate reports can be distressing. Volunteers and board members do not have access unless they are trained caseworkers. People who maintain the website do not read reports. Caseworkers working with children’s reports hold a current Working with Children Check.
The register records who viewed, coded, changed or exported each report. Caseworkers sign in with individual accounts and a six digit code from an authenticator app, which are switched on as each caseworker is onboarded.
How long we keep information
| Information | How long |
|---|---|
| Contact details (name, email, phone) | 7 years after the report is closed, or sooner if you ask, unless needed for a referral or legal matter still under way |
| Incident description and answers | 7 years, then de-identified so only statistical fields remain |
| Screenshots and photos | 7 years after the report is closed, or sooner if you ask |
| Caseworker notes | The same period as the report they relate to |
| De-identified statistics | Kept indefinitely so trends can be tracked over time |
| Technical logs kept by our host | Short-term only, as set by our hosting provider |
| Unsent drafts | Not stored by us. While you answer, your answers (never your contact details) are sent to our server, or to our AI reader when it is switched on, to suggest a category or read a screenshot, and are not kept afterwards. The draft itself stays in your browser until you submit |
The retention period runs from the date a report is closed, and is not restarted when a caseworker opens or reads it. A lead caseworker checks for reports due for de-identification at least every three months, including reports that have been open for longer than the retention period. When the period ends, a caseworker de-identifies the report so it can no longer reasonably identify anyone: names, contact details, free text, links, account names and attachments are removed, and the statistics and coding remain. A lead caseworker can also delete a report entirely. Both actions are recorded. You can ask us to delete your report sooner by contacting ggupta@saraswati.care.
If something goes wrong
We have a data breach response plan. If personal information is lost or accessed without permission, we act quickly to contain it and assess the risk. If it is likely to cause serious harm, we notify the people affected and the Office of the Australian Information Commissioner under the Notifiable Data Breaches scheme. We review every incident so it does not happen again.
Reviews
We review our security and this policy at least once a year, and before any major change, such as switching on AI screenshot reading. Before public launch we will complete a privacy impact assessment.