Every new account on an AI product costs real compute. Stop one person cycling accounts to run a business on your free tier.
What you will build: a check on the action you protect that allows normal users, asks doubtful ones to verify, and blocks the clear cases. See the ai free-tier abuse overview for the attack in pictures.
Before you start
Finish one Web quickstart (for example JavaScript or React) and one Server quickstart (for example Node.js). The decision code below is TypeScript; every server SDK returns the same event.
How the attack works
- Credit cycling. A new account when the free credits run out.
- Scripted sign-ups. Bots creating accounts to resell access.
- API key farming. Many keys, one machine, pooled usage.
What to look at
| Field | Why it matters here |
|---|---|
multiAccount | Accounts per device, across browsers. |
velocity | Many accounts from one visitor in a day. |
bot | Automated sign-up scripts. |
- 1
Identify on the action
Call the browser SDK when the user performs the action, not on page load, and send the
eventIdwith the request. Pass your own user id aslinkedIdwhenever you have one: the account signals count by it.TypeScriptconst { eventId } = await trace.get({ linkedId: user?.id, tags: { action: 'ai-free-tier-abuse' } }); - 2
Verify on your server
import { verifyEvent } from '@trace-id/sdk-node'; const event = await verifyEvent(eventId, { secretKey: process.env.TRACE_SECRET_KEY!, endpoint, timeoutMs: 1500 }).catch(() => null);try: event = client.get_event(event_id) except trace_device.VerifyError: event = None # fail opentry { $event = $client->getEvent($eventId); } catch (Trace\VerifyError $e) { $event = null; }Map<String, Object> event = null; try { event = trace.getEvent(eventId); } catch (Trace.VerifyException e) { /* fail open */ }const event = await trace.verify(eventId).catch(() => null);Events can be verified for 60 minutes. If verification fails, decide whether to fail open or closed; for most of this flow, fail open.
- 3
Decide
TypeScriptconst event = await trace.verify(eventId); if (event.signals.bot.detected) return block(); const credits = event.signals.multiAccount.detected ? 0 : FREE_CREDITS; await createAccount(email, { credits, deviceId: event.deviceId });trace.verify()is the embedded call; with a server SDK, use theeventfrom step 2 instead. The thresholds are yours. Trace scores and explains;event.reasonslists which signals added how much, so log it with every decision.
Example decisions
Free accounts from one device, illustrative:
| Account | Credits used | Decision | Why |
|---|---|---|---|
| [email protected] | 100 of 100 | Allow | First free tier |
| [email protected] | 100 of 100 | Allow | Second account, under the limit |
| [email protected] | new | Verify | multiAccount: ask for a card |
| [email protected] | new | Block | multiAccount + automatedInput |
Roll it out
- Ship steps 1 to 3 with logging only: record the decision, block nothing.
- After a week, look at who would have been blocked or asked to verify. Shared computers, offices and privacy tools are the usual false positives.
- Turn on Verify first, then Block for the clearest cases.