Skip to content
← All capabilities

Data protection & privacy

Privacy is not a bolt-on. Verifiable parental consent for minors, lawful-basis purpose tagging, data-principal rights with SLA clocks, and a dual-clock breach response - and student data is never sent to, or used to train, a third-party AI model.

Register your schoolBook a demo

See it live

Watch the real flow run

Built for the DPDP Act. Student data never trains an AI.

The payoff

Outcomes, not features

Verifiable parental consent for under-18 records
Per-purpose consent your parents actually understand
Data-principal rights (access, correction, erasure) on a clock
Breach response that meets CERT-In (6h) and DPDP (72h)

Start to finish

Inside data protection & privacy

The real end-to-end flow - from the school's first action to the audited, reason-coded outcome.

  1. 1
    Tag every purpose with a lawful basisNothing is processed on a vague blanket consent. Each purpose is bound to one of five DPDP lawful-basis tiers (statutory obligation, core-service, optional, voluntary, and safety), so a school can prove exactly WHY it holds each field. The two true consent tiers only turn on once a verified, unexpired consent record exists; the notice-only bases proceed but are still logged.
  2. 2
    Verify the parent for anything about a minorUnder-18 records trigger the Section 9 child guardrail. Processing is decided purpose-aware: a Fourth-Schedule child-interest purpose (attendance, assessment, health, child-safety) proceeds, a prohibited activity like behavioural tracking is blocked, and everything else falls back to full verifiable parental consent, backed by a signed DigiLocker age-token (DPDP Rule 10) rather than a self-declared checkbox.
  3. 3
    Encrypt the sensitive fields and log the accessAadhaar and other sensitive fields are held under AES-256-GCM field-level encryption, never as plaintext. Every read of a protected record is written to an append-only, tamper-evident audit trail (per-tenant sequence plus a keyed hash-chain), so who-saw-what is provable after the fact and cannot be quietly rewritten.
  4. 4
    Honour rights on a clock, and answer a breach on twoData-principal requests (access, correction, erasure) run on a configurable 30-day service target with a hard 90-day grievance ceiling. Erasure is a genuine crypto-erase: destroy the key and the ciphertext is unrecoverable forever. A reportable breach starts a dual clock at once, a 6-hour CERT-In technical report and a 72-hour Data Protection Board report, each tracked to its deadline.

The building blocks

The modules that do the work

The 4 modules that make up data protection & privacy.

Consent & privacy

Lawful-basis tiers, verifiable consent, Rule-3 notice.

Consent verification

Adult / age verification, DigiLocker-ready.

Sensitive data

Field-level encryption and governance.

Security audit

Append-only, tamper-evident audit trails.

Signature automations

What most systems don't do

Built-in mechanics you won't find elsewhere - running in the product, not drawn on a slide.

Five lawful-basis tiers with a purpose-aware child guardrail

Consent is not one on/off switch. Five explicit DPDP bases classify every purpose, and processing of a minor is decided per-purpose under Section 9(3): exempt child-interest purposes proceed, prohibited activities (tracking, behavioural monitoring, profiling, targeted advertising) are blocked, and anything else requires verifiable parental consent. Fail-safe: an unparseable date of birth is treated as a child.

Verifiable parental consent via a signed DigiLocker age-token

DPDP Rule 10 wants VERIFIABLE consent, not a checkbox. Vidyom verifies a signed, expiring DigiLocker age-token before writing gold-tier parental evidence. The token payload carries only an opaque pairwise subject reference and an adult boolean, never a name or DOB, and the raw token and secret are never logged. Off the live credential it fails closed rather than trusting the public dev stub.

AES-256-GCM field encryption with a true crypto-erase

Sensitive fields are sealed with AES-256-GCM, and erasure is not a soft delete. Destroying the per-record key makes the ciphertext fail its GCM auth tag forever, an unrecoverable NIST SP 800-88 crypto-erase that satisfies the DPDP right-to-erasure for one token or an entire subject vault, with no plaintext fallback on a decrypt failure.

Tamper-evident audit chain: keyed hash-chain, Merkle head, external witness

Access logs are not merely append-only by privilege; each row folds a keyed HMAC-SHA256 hash-chain (prev-hash to row-hash) over a canonical payload, so a privileged insider who mutates or deletes a middle row is caught on verify. A Merkle tree head can be published to an independent external witness to catch tail-truncation too, wired flag-off with an honest no-egress stub until a live timestamping service is configured.

Also built in

More in data protection & privacy

Capabilities that already ship in the product - each shown live, not just named.

Access governance

Custom roles, recertified & dual-controlled

Build your school's own roles by toggling per-module viewer, editor and admin access. Recertify who holds each role on a schedule, and require two-person approval before any sensitive money action goes through. Every grant is server-enforced and audited.

  • Compose a role like "Accountant" from per-module permission toggles
  • Periodic access recertification with a signed, dated stamp
  • Dual-control: 2-of-2 approval for Fees & Payroll writes

POCSO Section-23

A protected child’s identity is redacted in every view

Flag a child victim as confidential once, and their name, photo and contact are automatically masked across every screen, list, report and export — the identity can never quietly re-surface.

  • One canonical redaction choke-point — fail-closed, so an unclassified record is never leaked by accident
  • A reveal needs the explicit pocso:reveal capability plus step-up re-auth
  • Every reveal is a logged break-glass, then the record re-seals to redacted

Straight answers

Data protection & privacy - common questions

Is student data ever sent to an AI model?

No. Student PII is never sent to, or used to train, any third-party model. The AI-assist paths are deterministic templates by default, and a runtime guard refuses any prompt that even carries a name-like or contact field key. The privacy posture and the intelligence posture are the same promise.

When a parent asks you to delete a child, is it really gone?

Yes, and provably. Encrypted fields are crypto-erased by destroying the key, which permanently breaks decryption of the ciphertext, and the request runs on a 30-day target with a 90-day grievance ceiling. It is a real erasure, not a hidden flag.

What actually happens in a data breach?

A dual clock starts immediately: a 6-hour CERT-In technical report and a 72-hour Data Protection Board report, each tracked to its own deadline so neither statutory window is missed. The obligation is durable, so a crash between raising and reporting is reaped rather than dropped.