All articles Fraud & ATO

Edtech Account Sharing: Device Signals for Exam Integrity

Dark minimal scene of a laptop and a phone sending blue light beams toward one shared glass token, representing one account used from two devices.

In most edtech products, edtech account sharing does not look like fraud at first. A school buys a block of seats, a student texts the login to a classmate before a quiz, and a family shares one subscription across two laptops and a phone. Nothing in the password field looks wrong, because nothing is wrong with the password: it is the right one, used by the wrong person.

The cost lands in two places: seat revenue, because one paid license serves several learners, and assessment integrity, which is harder to repair when a graded exam is completed by someone other than the enrolled student. Both come from the same root cause: an account with no reliable link to a device.

This guide covers which device signals expose a shared login, and how to turn them into a login policy that allows, steps up or blocks.

Why “one account, one student” is hard to enforce

Password sharing is a social habit, not a technical exploit. Students trade credentials the way they trade notes, and legitimate users make the picture noisier: family laptops, library machines, campus NAT where hundreds of devices leave through one address, mobile carriers that rotate addresses constantly. That is why IP rules fail in both directions: they flag a lecture hall as abuse and miss two roommates on the same Wi-Fi using different machines. Cookies are no better, since clearing storage or opening an incognito window resets the counter.

The seat problem and the integrity problem need different policies

A shared login during a practice quiz is a billing question. The same login inside a proctored exam is an academic integrity question. Use one signal set with two policies: meter or upsell seat sharing in practice content, and hold a much stricter line once an assessment window opens.

Device signals that carry the most weight

Two laptop silhouettes at night sharing one blue connection line, illustrating a shared login on separate devices.

Fingerprint stability and device diversity per account

The strongest pattern is not where a login comes from but how many distinct, stable devices one account uses. A student with one laptop and one phone produces a stable fingerprint all term. Many stable fingerprints on one account in a week is a shared credential, not a jet-setting learner. Device fingerprinting (https://pryntid.com/device-fingerprinting) is the layer that gives you that count, and Prynt Identify exposes it next to bot, VPN and tampering signals.

Network signals as context, not verdict

VPN, proxy, Tor and datacenter detection tells you how a connection is routed, not who is on it. A datacenter address during an exam deserves attention; a mobile carrier address on a Tuesday night does not. Read network signals next to the device and the account’s own history.

Tampering and automation

Assessment abuse increasingly involves automation: scripted browsers, spoofed fingerprints, answer-farm tooling. Tampering signals plus AI-agent and bot detection help you separate a student on a slow laptop from a headless browser answering at machine speed.

Turn those signals into a login decision

Signals are useless until they produce an action at login. The login use case in Prynt Control returns allow, step-up or block, with a reason code and a suggested step such as a passkey, email verification or a captcha, plus a deviceTrust value of trusted, new or risky.

// pseudo-code: your server makes the call, Control returns the decision
const decision = await control.decide({
  useCase: "login",
  userId: account.id,
  requestId: req.id,
});

if (decision.action === "step-up") return requireStep(decision.suggestedStep);
if (decision.action === "block") return deny(decision.reason);
// otherwise: allow

Control never blocks on its own; your code decides what to do with the answer. When a second concurrent session appears, the extra_seat use case returns a separate decision, so you can offer an upgrade instead of locking out a paying learner.

Make the step-up cheap and specific

Most sharing is not malicious. Verify email or a passkey on the second device, keep the session alive and let the pattern accumulate. Save blocks for repeated risky devices, tampered fingerprints or a new device appearing after the exam has started.

Mirror the policy before you enforce it

Control starts with a 14-day mirror period: decisions are evaluated and logged, nothing is blocked. Read that log first. Check which accounts would have been stepped up, look for campus networks and shared devices in the list, and only then switch enforcement on. That review is the difference between a control that holds and a support queue full of locked-out students.

Assessment integrity: watch the session, not the password

Once an exam starts, the question changes from “is this device known?” to “is it still the same device?” A fingerprint swap, a mid-exam network change or a second device joining the account is a reason to flag the attempt for review, not to fail it automatically. Attach the reason code, keep the evidence and route it to the instructor. The same logic applies to free practice content, the usual entry point for re-sharing: a meter that counts per device and per account, and that incognito does not reset, closes the gap cookie counters leave open.

Abstract network of device nodes clustering into one account, showing device diversity as a sharing signal.

Privacy boundaries to state clearly

Tell schools and students which signals you use and why, keep only what a decision needs, and store identifiers as keyed hashes; phone numbers and cards are kept that way in Control. Retention and processing terms belong with the school, and none of this is legal advice. Implementation details are in the docs: https://pryntid.com/docs.

FAQ

Three translucent panels in dark space showing a policy evaluated in stages before a decision is enforced.

Is a shared login the same as account takeover?

No. Sharing is voluntary and usually happens between people who know each other; takeover is credential theft. The signals overlap, which is why login policy and account takeover controls (https://pryntid.com/account-takeover) are worth designing together.

Do campus VPNs break device fingerprinting?

They blur IP rules far more than device fingerprints. A campus network hides who is behind an address, but each device still presents its own fingerprint, so device diversity on the account stays readable.

Can we just cap concurrent sessions?

That catches live sharing and misses the common pattern: one student finishes, the next logs in an hour later. A cap is a useful ceiling; device-level policy is what tells you an account has more than one person behind it.

You cannot enforce a policy you cannot see. Log login decisions for two weeks, compare device diversity per account with your seat count, and choose the step-up you are willing to hand a student at 9 a.m. on exam day. Prynt Control gives you the decision layer, the reason codes and a mirror period to test the policy before it blocks anyone: https://pryntid.com.

Try it free

Prynt is device intelligence with a free plan: 5,000 identifications a month, visitor IDs with confidence and bot detection — no credit card. Upgrade to Pro for VPN, proxy, tampering and behavioral signals.

Keep reading