iGamingTask Prompt

Task Prompt to Design a Wager-Time Facial Age Verification Gate

This **Arc** task prompt takes a sportsbook's available age signals and the pending federal **facial age verification mandate** for sportsbooks and prediction markets, then designs the **wager-time age gate**. Arc, the deepidv credential gateway, ingests and ranks every per-session age signal, **facial age estimation** with passive liveness from deepeye, **OS-level age assurance** signals, **eIDAS 2.0 / mDL / ZKP age tokens**, and the account's verified age of record, builds the decision matrix for which signal combinations clear a wager instantly versus trigger a fresh camera-session estimation versus hard-block, enforces the two mandate trigger points at login and wager placement inside the client-edge latency budget, and applies privacy constraints so no biometric template is stored. Built for age and trust leads at US-licensed operators who must catch the borrowed-account case, a minor using an adult's authenticated session, before the mandate lands.

How to use this prompt

1
Open Arc in the deepidv dashboard and paste the full prompt, or run it in Claude, ChatGPT, or Gemini if you are drafting the age-gate design outside the platform.
2
Replace the INPUT section with the age signals your platform can access today, your current login and wager flow, and your device-attestation coverage.
3
Run the prompt and read the decision matrix first: it maps which signal combinations clear a wager, which trigger a fresh estimation, and which hard-block.
4
Hand the integration spec, endpoints, decision codes, retry logic, and audit-log schema, to your platform engineers, and route the privacy constraints to compliance.
5
Re-run the prompt when the federal bill advances or a state adds an age rule so the gate stays aligned before enforcement begins.

The prompt

Arc, design a wager-time age assurance gate for a US-licensed online sportsbook preparing for the pending federal facial age verification mandate covering sportsbooks and prediction markets.

Requirements:
1. Ingest and rank every age signal available per session: facial age estimation with passive liveness from deepeye, OS-level age assurance signals where the platform provides them, eIDAS 2.0 / mDL / ZKP age tokens where the user holds one, and the account's verified age of record.
2. Build the decision matrix: which signal combinations clear a wager instantly, which trigger a fresh camera-session estimation, and which hard-block pending full re-verification.
3. Enforce the two mandate trigger points: login and wager placement. Wager-time checks must complete inside the client-edge latency budget; specify the fallback when a device cannot attest.
4. Apply privacy constraints: no stored biometric templates, age band outcomes only, evidence trail limited to verdict, signal sources, and timestamps.
5. Output an integration spec for my flow: endpoints, decision codes, retry logic, and the audit log schema a state regulator or a future federal examiner would request.

Assume borrowed-account fraud is the primary threat: a minor using an adult's authenticated session. Design every branch to catch that case.

Test it in Claude or another LLM

This prompt is built for the Arc agent inside deepidv, where Arc ranks a session's live age signals and returns a wager-time decision. You can dry-run the same workflow in any general LLM first with synthetic signal data to see the decision-matrix shape before wiring it to live traffic.

1
Paste the full prompt into Claude, ChatGPT, or Gemini, but replace the opening 'Arc,' with a role instruction such as 'Act as an age-assurance architect designing a wager-time age gate for a US-licensed sportsbook.' Keep the OUTPUT sections exactly as written.
2
Under the INPUT section, paste the synthetic sample block below so the model has the available signals, the flow, and the attestation coverage to design against.
3
Add one framing line: 'This is synthetic test data. Assume borrowed-account fraud is the primary threat and design every branch to catch it.'
4
Check the output shape: a ranked signal list, a decision matrix for clear/re-estimate/block, the login and wager trigger enforcement with a fallback for un-attestable devices, the privacy constraints, and an integration spec with an audit-log schema. If the design stores a biometric template anywhere, tighten the privacy constraint line and re-run.
5
Once the output shape is right, run it live in the deepidv dashboard where Arc ranks your real session signals.

Synthetic sample data to paste alongside the prompt

Fake test data, safe to share with any LLM. Swap in your own once the output looks right.

AVAILABLE AGE SIGNALS (synthetic, fake):
- deepeye facial age estimation + passive liveness (available)
- OS-level age signal (available on ~40% of sessions, fake)
- eIDAS 2.0 / mDL / ZKP age token (available on ~5% of sessions, fake)
- account verified age of record (available for returning users)
FLOW (fake): login screen, then bet-slip confirmation as the wager trigger
ATTESTATION COVERAGE (fake): 82% of devices can attest; 18% cannot
LATENCY BUDGET (fake): wager-time decision hard cap 150ms
THREAT (fake): minor using a parent's logged-in session
OPEN ITEM (fake): fallback path when a device cannot attest and no token is present

Pairs with on deepidv

Sources & further reading

FAQ

What is a wager-time age gate?

A wager-time age gate is a session-level control that confirms a live adult is present at both login and the moment a bet is placed, rather than only at account signup. It ranks available age signals, facial age estimation over passive liveness, OS-level age assurance, and ZKP age tokens, and decides whether to clear, re-estimate, or block, all inside the client-edge latency budget.

Why check age at wager placement and not just signup?

Onboarding age checks verify the account holder once. They do not catch a minor using an adult's already-authenticated session, which is the primary fraud path the pending mandate targets. A wager-time check re-confirms a live adult is present at the moment of the bet, closing the borrowed-account gap.

Can this age gate run without storing biometric data?

Yes. The design constraint is age-band outcomes only, with the evidence trail limited to verdict, signal sources, and timestamps. Facial age estimation over passive liveness issues an age band without persisting a biometric template, which matches the privacy-preserving framing in the pending House bill and state biometric laws.

Run it with live verification data

These prompts work in any LLM. Inside the deepidv dashboard, Luna, Arbiter, and Arc run them against your real sessions, screening lists, and audit trails.

Book a Demo