Responsible-gambling intervention. Promo eligibility. KYC gate. Withdrawal decision. Prediction-market trade. Model-driven line move. Right now every one of those is a self-written log. In 2026, self-written logs are the exact thing the UK Gambling Commission, US state regulators, and plaintiffs' counsel are refusing to treat as evidence. Below is a working map of how Hive turns each of those actions into a receipt that FanDuel didn't sign alone — with a proof boundary that never over-claims.
Regulators, courts, the AI question, and integrity investigations are climbing at the same time — and all four demand the same primitive: independent, signed, contemporaneous proof that the intervention or decision happened, when, and on what basis. A self-written log is a document written by the defendant.
Each row is a real 2026 pain surface, the specific proof gap it opens, and where a Hive receipt would sit on the rail. Nothing here claims FanDuel currently lacks these capabilities in code — the claim is that independent, signed, contemporaneous proof of those capabilities firing is what regulators, courts, and integrity partners are now asking for.
| Exposed surface | What a Hive receipt proves | Rail window |
|---|---|---|
| Responsible-gambling intervention — deposit-limit prompt, cool-off suggestion, self-exclusion offer, marker-of-harm alert | The intervention actually rendered to that account, at that timestamp, under that policy version, with that model score or rule trigger. Signed under did:hive:hivemorph. Later reproducible in front of a regulator or an examiner. |
pre-broadcast stoppable |
| Promo eligibility & deceptive-design defense — daily bonus bet, parlay boost, VIP touch, in-app nudge | The exact eligibility criteria evaluated for the user, the resulting decision, and the promo copy version served. Turns a plaintiffs' subpoena for “every in-app promo shown” into a signed, exportable evidence bundle. | observed |
| KYC / affordability gate — verification hold, source-of-funds request, withdrawal delay | A receipt that the gate fired at the required threshold, with the required inputs, and either passed or held — with the reason. Turns 24-48h KYC withdrawal-delay complaints into a signed record of why the hold existed. | pre-broadcast stoppable |
| Prediction-market trade — order placement, matching, settlement, dispute | Every trade attested at intent, at match, and at settlement. Signed proof of the eligibility check, the counterparty match rules, and the settlement basis — the shape prediction-market regulators (CFTC-adjacent) are already asking about. | submitted → chain-verified |
| Model-driven line move / limit adjustment — automated price change, exposure cap, sharp-account throttle | Which model version priced the market at that moment, on which inputs, resulting in which decision. Turns an integrity investigation into a re-playable, signed trail — not a data-team screenshot. | observed |
| Suspicious-activity / AML flag — structuring, unusual pattern, watchlist match | A signed receipt that the alert fired, with a hash of the input feature vector and the model output. Suitable for SAR export and for defending the alert-that-didn't-fire question after the fact. | evidence-only final |
| Agent / LLM-assisted customer action — chatbot dispute, self-service limit change, AI concierge recommendation | Every AI-mediated action independently receipted — what the model saw, what it recommended, what the user accepted. Directly answers the “are your LLMs acting as offshore-casino concierges?” question with a signed no. | pre-broadcast stoppable |
RG prompts, promo gates, KYC holds, trades, line moves, AML flags, AI-mediated actions — streaming past like the odds board, each stamped with its proof boundary. The receipt never over-claims which window applies.
Nothing here asks FanDuel to change how it prices markets, runs its RG program, or evaluates promos. The primitives sit alongside those systems and receipt what they already do.
Every consequential action — RG prompt, KYC gate, promo eligibility, trade, alert — passes through POST /v1/receipt-relay/event and comes back Ed25519-signed with the proof boundary made explicit: self-attested, Hive-observed, submitted, or chain-verified. Sign-and-return is ~70 ms median in production, so the receipt path never becomes a bottleneck on the play path.
Receipts belong to signing groups — per player, per market, per promo campaign, per compliance case. A group is queryable, exportable as a signed bundle, and independently verifiable by any third party against the Hive public key. This is the substrate a regulator or plaintiffs' subpoena team can re-verify offline, with no FanDuel-provided software in the loop.
R3Pv reads a receipt group and emits a signed vector over verification depth, weakest proof boundary, healing state, recoverability window, routing recommendation, and permitted next actions. For an RG case: proof depth on the last three interventions, weakest boundary in the trail, the recoverability window, and what the risk stack is allowed to do next. Live at POST /v1/r3pv/vector.
Protected Flow reads the vector, applies a policy pack (default, strict_bank, market_integrity, or a FanDuel-specific pack), and emits a signed decision: permit, permit_with_evidence, require_approval, hold, recover, reroute, escalate, or block. Strictest wins — policy can only make decisions stricter, never looser. Live at POST /v1/protected-flow/assess.
Same receipt trail, different policy pack, stricter decision. The needle can ratchet up — it can never be talked back down. That's not a promise, it's the mechanism: strictest wins.
Drag the handle from “still stoppable” to “evidence-only final” and watch the window physically close. Hive records which window applied at every step — and never markets a recoverability the rail doesn't support. That's the exact posture the UKGC asked for.
The receipt never over-claims which window applies.
The demo below sends a synthetic RG-intervention receipt and gets back an Ed25519-signed proof. Nothing in this request touches real player data — the shape is illustrative. Anyone inside FanDuel can run this on their laptop in under a minute.
Confirms the relay is live and returns the key ID that will sign every receipt.
curl https://receipts.thehiveryiq.com/v1/receipt-relay/health
Illustrative shape. No player data, no FanDuel systems touched. Response includes receipt_id, payload_sha256, sig_b64u, algorithm: Ed25519, key_id: did:hive:hivemorph.
# Sign an RG intervention receipt (illustrative — no player data) curl -X POST https://receipts.thehiveryiq.com/v1/receipt-relay/event \ -H "Content-Type: application/json" \ -d '{ "external_system":"fanduel_illustrative", "action":"rg_intervention_shown", "subject":"deposit_limit_prompt", "actor":{"type":"policy_engine","id":"rg-marker-of-harm-v3"}, "intent":{"user_ref":"account_placeholder_0001","intervention":"deposit_limit_prompt","policy_version":"rg-2026-06","trigger":"velocity_score_threshold"}, "evidence":[ {"mode":"observed","note":"prompt rendered client-side and dismissed within 2s"} ] }'
Same receipt trail, stricter policy pack → stricter decision. The response is a signed assessment that round-trips through POST /v1/receipt/verify.
curl -X POST https://receipts.thehiveryiq.com/v1/protected-flow/assess \ -H "Content-Type: application/json" \ -d '{ "receipts":[/* rg receipt from step 2 */], "policy_pack":"strict_bank", "declared_exposure_usd": 250.00 }'
A full signing group — every receipt, the signed vector, the assessment, the healing trail — packaged as a single signed bundle. Suitable for regulator export, subpoena response, or internal integrity review. Independently verifiable offline against the Hive public key.
curl https://receipts.thehiveryiq.com/v1/protected-flow/evidence/{group_id}/export
The lever above is the same taxonomy in words. Each state names the window that actually exists — and the receipt records which window applied at each step, so the after-the-fact reconstruction is signed, not narrated.
The most useful thing Hive can do in a company under internal pressure is not to arrive as a marketing partnership. It's to arrive as an infrastructure layer that turns an internal review into a report the champion forwards — without defending a new vendor relationship first.
RG interventions on top-decile-velocity accounts, or KYC holds on flagged withdrawals. Async through Receipt Relay. Zero change to the surface itself. The signed trail exists from day one.
Receipts bundled per case. R3Pv emits the signed decision vector — proof depth, weakest boundary, recoverability window, permitted next actions. Reproducible in front of a regulator.
FanDuel-tuned policy pack. Signed assessments. One exportable evidence bundle — receipts, vector, decision, healing trail — verifiable offline against the Hive public key.
Sportsbooks have programmable pricing, programmable promos, programmable RG. What no US sportsbook has yet is a signed, third-party-verifiable proof trail across all of them at once. The category is not “more compliance software.” It's independent proof-of-action — the layer regulators, courts, integrity partners, and the market are all converging on at the same time.