Prompt injection in iGaming and sports betting AI

Four adversarial injection surfaces in iGaming and sports betting AI

1. Age and identity verification document image injection (COPPA 15 USC §6501, state Gaming Control Board requirements)

Age and identity verification KYC AI processes government-issued ID document scan images displaying driver’s licence or passport biographical data page with name, date of birth, address, and document number fields, age verification facial comparison display images showing AI-matched customer selfie against extracted government ID photograph with liveness detection confidence score display, KYC document classification display images showing AI-assigned document authenticity score, country of issuance verification status, and security feature integrity assessment indicators, geolocation compliance indicator display images showing customer’s IP address geolocation comparison against licensed jurisdiction boundary maps with approval or restriction status determination, and OFAC sanctions screening result display images showing player name, address, and date of birth comparison against OFAC SDN List with match confidence score from DraftKings AI at 5.7 million monthly active users across 23 licensed US states processing government ID scan and facial comparison display images through AI-assisted KYC onboarding, age verification, and geolocation compliance tools; FanDuel Sportsbook AI at 8.5 million daily active users across 22 US states and Canada processing government ID document scan, KYC classification, and OFAC screening display images through AI-assisted account onboarding and regulatory compliance tools; and BetMGM AI at 25 US states and £1 billion or more annual UK revenue processing government ID scan, facial comparison, geolocation, and OFAC screening display images through AI-assisted KYC regulatory compliance and AML customer due diligence tools — extracting age verification eligibility determinations, KYC document authenticity assessments, geolocation jurisdiction compliance approvals, and OFAC sanctions screening pass/fail results from government ID document scan and identity verification display image inputs in AI-assisted online gaming and sports betting regulatory KYC onboarding pipelines at account onboarding volumes that make individual human KYC reviewer examination of every AI-processed identity document impracticable for large-scale online gaming platform operations.

The adversarial injection surface is the government ID document scan image, age verification facial comparison display image, or KYC document classification display image submission pathway: DraftKings AI, FanDuel AI, or BetMGM AI identity verification document scan and KYC display images submitted through AI-assisted age verification and regulatory compliance onboarding tools for AI KYC determination record generation and state Gaming Control Board compliance filing. An adversarially crafted government ID document scan image — in which pixel perturbations applied to the date of birth field display region, the document authenticity indicator overlay, the facial photograph matching confidence score display, or the geolocation jurisdiction boundary comparison display cause the AI to classify an under-18 applicant’s government ID as belonging to an eligible adult meeting state gaming age verification requirements, or to classify a document from a self-excluded or sanctioned player as an authentic eligible customer document — can enable account creation and betting access for under-age applicants, suppress self-exclusion registry match indicators that would otherwise prevent excluded players from accessing gaming accounts, enable OFAC-sanctioned individuals to create gaming accounts, or facilitate geolocation bypass by players in non-licensed jurisdictions. In online gaming onboarding where DraftKings AI or FanDuel AI processes thousands of KYC document images per day without individual human reviewer examination of every AI-processed identity document before the AI eligibility determination governs account creation authorisation, adversarial injection in KYC document images creates state Gaming Control Board age verification, COPPA 15 USC §6501, OFAC 31 CFR Part 500, and FinCEN CDD Rule 31 CFR §1010.230 dimensions.

The state Gaming Control Board requirements, COPPA 15 USC §6501, OFAC 31 CFR Part 500, and FinCEN CDD Rule 31 CFR §1010.230 regulatory consequences of adversarially corrupted KYC age verification classification span Nevada Gaming Control Board Regulation 5A requiring online gaming operators to verify customer age and identity before permitting account funding and wagering — adversarially bypassed age verification AI creates NGC Reg 5A compliance violation dimensions with licence revocation potential; New Jersey Division of Gaming Enforcement N.J.A.C. 13:69C online gaming age verification requirements establishing that New Jersey licensed operators must verify that customers are 21 years of age or older before permitting account wagering — adversarially corrupted age verification AI creates NJDGE N.J.A.C. 13:69C compliance violation and civil penalty dimensions; Pennsylvania Gaming Control Board 58 Pa. Code §401a online gaming registration requirements establishing age and identity verification standards — adversarially bypassed PA PGCB age verification creates licence compliance violation dimensions; COPPA 15 USC §6501 protections for children’s privacy and the broader principle that gaming operators are legally prohibited from accepting wagers from minors under state age verification statutes — adversarially bypassed under-18 gaming AI creates minor gambling exposure dimensions; OFAC 31 CFR Part 500 SDN List screening requirements establishing that US persons including gaming operators must not transact with OFAC-designated individuals — adversarially corrupted OFAC screening AI creates OFAC civil penalty exposure up to $1.3 million per transaction; and FinCEN CDD Rule 31 CFR §1010.230 beneficial ownership and customer risk rating requirements applicable to gaming industry AML programmes. Threshold: 70 for age and identity verification document image injection — reflecting state Gaming Control Board age verification, COPPA 15 USC §6501, OFAC SDN 31 CFR Part 500, and FinCEN CDD 31 CFR §1010.230 regulatory dimensions.

2. Odds display and line movement image injection (Wire Act 18 USC §1084, IGRA 25 USC §2702)

Odds display and line movement integrity AI processes live odds display dashboard images showing moneyline, spread, and totals pricing fields for active sports betting markets with AI-readable line movement magnitude and velocity indicators, in-play betting market display images showing real-time odds updates with temporal sequence and price movement direction indicators during active sports events, line movement indicator chart images displaying historical odds trajectory visualisations for pre-game and live sports betting markets with sharp money indicator overlays and opening-to-current line movement magnitude displays, arbitrage detection display images showing cross-book odds comparison matrices and value bet probability threshold indicators, and suspicious trading pattern alert display images showing AI-generated market integrity anomaly scores and betting pattern clustering indicators from Kambi Sports Solutions AI at 50 or more sportsbook operator clients and 30 or more global markets processing live odds display and line movement images through AI-assisted odds integrity monitoring, arbitrage detection, and market manipulation screening tools; Sportradar AI trading platform at 1,000 or more sports events per day and 900 or more sports organisation clients processing real-time odds display and suspicious trading alert images through AI-assisted integrity monitoring, match-fixing detection, and sports betting data integrity tools; and IGT Advantage AI at 100 or more jurisdictions and 400 or more operator clients processing odds display and market monitoring images through AI-assisted gaming system odds management and integrity compliance tools — extracting Wire Act interstate wagering compliance determinations, IGRA tribal gaming compact integrity assessments, sports betting market manipulation flags, and cross-book arbitrage opportunity detections from odds display dashboard and line movement chart image inputs in AI-assisted sports betting integrity monitoring pipelines.

The adversarial injection surface is the live odds display dashboard image or line movement indicator chart image submission pathway: Kambi AI, Sportradar AI, or IGT Advantage AI odds display and market integrity monitoring images submitted through AI-assisted odds integrity screening and market manipulation detection tools for AI integrity determination record generation and gaming regulator reporting. An adversarially crafted live odds display dashboard image — in which pixel perturbations applied to the moneyline price display fields, the line movement magnitude and velocity indicator overlays, the sharp money indicator displays, or the arbitrage opportunity comparison matrix cells cause the AI to classify an in-play betting market with coordinated suspicious betting patterns — consistent with match-fixing, insider information trading, or cross-book arbitrage exploitation — as a clean market with no integrity anomalies and no suspicious trading activity when the actual real-time odds movement data evidences patterns meeting Sportradar AI’s integrity alert escalation thresholds — can suppress a sports betting integrity anomaly that would otherwise generate a Sportradar Integrity Services escalation to the relevant sports governing body, a Wire Act compliance review for interstate wagering pattern anomalies, an IGRA tribal gaming compact integrity notification, or an FBI sports betting integrity programme referral. In sports betting operations where Kambi AI or Sportradar AI processes thousands of odds display images per day across hundreds of live sports markets without individual human trading desk analyst review of every AI-processed odds display before the AI integrity determination governs the market escalation decision, adversarial suppression of sports betting integrity indicators creates Wire Act 18 USC §1084, IGRA 25 USC §2702, and PASPA-successor state sports betting statute dimensions.

The Wire Act 18 USC §1084, IGRA 25 USC §2702, PASPA-successor state sports betting integrity statutes, and FBI integrity programme regulatory consequences of adversarially suppressed odds display and line movement classification span Wire Act 18 USC §1084 interstate wagering prohibitions establishing criminal penalties of up to 2 years imprisonment for use of wire communication facilities in connection with bets or wagers placed in interstate commerce — adversarially corrupted odds display AI that suppresses interstate suspicious betting pattern indicators affecting Wire Act compliance monitoring creates Wire Act §1084 dimensions; IGRA 25 USC §2702 tribal gaming purpose requirements and associated NIGC gaming ordinance provisions establishing that gaming operations must maintain integrity standards consistent with tribal gaming compact requirements — adversarially corrupted AI integrity monitoring for tribal gaming online operations creates IGRA and NIGC compliance dimensions; PASPA-successor state sports betting integrity requirements in NJ, PA, NV, and other regulated markets establishing that licensed sports wagering operators must implement market monitoring and integrity reporting requirements — adversarially suppressed AI integrity alerts create state gaming regulator reporting obligation failure dimensions; and FBI sports betting integrity programme coordination mechanisms that rely in part on commercial AI integrity monitoring platforms including Sportradar AI for real-time suspicious betting activity notification — adversarially corrupted Sportradar AI integrity output creates FBI programme coordination failure dimensions. Scientific Games AI lottery system integrity monitoring creates additional IGRA and state lottery commission integrity requirement dimensions for lottery-linked sports betting products. Threshold: 55 for odds display and line movement image injection — reflecting Wire Act 18 USC §1084 interstate wagering compliance, IGRA 25 USC §2702 tribal gaming compact integrity, PASPA-successor state sports betting statute reporting, and FBI integrity programme coordination dimensions.

3. AML and suspicious transaction documentation injection (BSA 31 USC §5318, OFAC SDN 31 CFR Part 500)

AML and suspicious transaction documentation AI processes transaction monitoring alert display images showing AI-generated risk scoring for individual account transactions, deposit and withdrawal pattern anomaly indicators, high-risk jurisdiction transaction flag indicators, and velocity-based transaction clustering alert displays, currency transaction report (CTR) documentation display images showing transaction amounts exceeding $10,000 CTR threshold, beneficiary information, and structured transaction avoidance indicator fields with AI-assisted BSA filing status determination overlays, suspicious activity report (SAR) filing decision display images showing AI-generated SAR narrative summary, activity category classification (layering, structuring, smurfing, integration), and filing recommendation status with supporting transaction evidence links, high-risk player alert indicator display images showing player risk tier classification, PEP (politically exposed person) indicator status, adverse media screening results, and enhanced due diligence requirement triggers, and OFAC SDN matching status display images from DraftKings AI at 5.7 million monthly active users processing transaction monitoring alert and SAR documentation display images through AI-assisted BSA/AML compliance, CTR filing, and OFAC screening tools; FanDuel AI at 8.5 million daily active users processing AML transaction alert and high-risk player indicator display images through AI-assisted BSA programme and FinCEN compliance tools; and BetMGM AI at 25 US states and UK operations processing AML transaction monitoring, SAR documentation, and OFAC screening display images through AI-assisted US BSA and UK Proceeds of Crime Act AML compliance tools — extracting FinCEN CTR filing obligations, BSA SAR filing determinations, OFAC sanctions match escalation decisions, and high-risk customer enhanced due diligence triggers from AML transaction monitoring alert and SAR documentation display image inputs in AI-assisted online gaming AML compliance pipelines.

The adversarial injection surface is the transaction monitoring alert display image, CTR documentation display image, or SAR filing decision display image submission pathway: DraftKings AI, FanDuel AI, or BetMGM AI AML transaction monitoring and SAR documentation display images submitted through AI-assisted BSA compliance and FinCEN reporting tools for AI SAR filing determination record generation and FinCEN filing. An adversarially crafted transaction monitoring alert display image — in which pixel perturbations applied to the transaction risk score numerical display, the structuring pattern indicator overlay, the high-risk jurisdiction flag display, or the SAR filing recommendation status indicator cause the AI to classify a player account with FinCEN-reportable suspicious transaction activity — including structured deposits below $10,000 designed to avoid CTR reporting thresholds, rapid deposit-and-withdrawal cycling patterns consistent with layering, or large cash equivalent transactions from high-risk jurisdictions — as a clean account with no BSA reporting obligations when the actual transaction monitoring data evidences activity meeting FinCEN’s SAR filing guidance thresholds for online gaming platforms — can suppress a BSA reporting obligation that would otherwise generate a FinCEN CTR filing, a SAR filing referral to FinCEN, an OFAC SDN match escalation, or an enhanced due diligence trigger for a PEP or high-risk player account. In online gaming AML programmes where DraftKings AI or FanDuel AI processes thousands of transaction monitoring alerts per day without individual human BSA compliance officer examination of every AI-processed alert display before the AI determination governs the SAR filing decision, adversarial suppression of BSA indicators creates FinCEN BSA 31 USC §5318, FinCEN CDD 31 CFR §1010.230, and OFAC 31 CFR Part 500 regulatory dimensions.

The FinCEN BSA 31 USC §5318, FinCEN CDD Rule 31 CFR §1010.230, OFAC SDN 31 CFR Part 500, and FinCEN 2021 gaming AML guidance regulatory consequences of adversarially suppressed AML transaction monitoring classification span FinCEN BSA 31 USC §5318 anti-money laundering programme requirements establishing that financial institutions including casinos and online gaming platforms must implement internal controls, independent testing, training, and suspicious activity reporting as components of a written AML programme — adversarially corrupted AI transaction monitoring that suppresses SAR filing obligation triggers creates BSA §5318 AML programme failure dimensions with civil money penalty exposure up to $1 million per day for wilful violations; FinCEN CDD Rule 31 CFR §1010.230 customer due diligence requirements establishing beneficial ownership identification and customer risk rating requirements applicable to gaming platform customers — adversarially suppressed high-risk player and PEP indicator AI creates CDD Rule enhanced due diligence programme failure dimensions; OFAC 31 CFR Part 500 SDN List restrictions establishing that US persons must not transact with OFAC-designated individuals — adversarially corrupted OFAC SDN screening AI that suppresses match indicators for designated persons creates OFAC civil penalty exposure up to $1.3 million per transaction; FinCEN’s 2021 guidance on AML programme requirements for gaming industry including online sports wagering platforms establishing SAR filing thresholds and structuring detection requirements specific to online gaming transaction patterns; and UK Proceeds of Crime Act 2002 §330 failure-to-disclose obligations applicable to BetMGM UK operations where adversarially suppressed AML transaction AI creates POCA §330 criminal liability dimensions for failure to report known or suspected money laundering. Threshold: 65 for AML and suspicious transaction documentation injection — reflecting FinCEN BSA 31 USC §5318 SAR filing obligations, FinCEN CDD 31 CFR §1010.230 customer due diligence, OFAC SDN 31 CFR Part 500 sanctions screening, and FinCEN 2021 gaming AML guidance dimensions.

4. Responsible gambling problem indicator image injection (NCPG standards, GAMSTOP UK self-exclusion)

Responsible gambling problem indicator AI processes problem gambling risk score display images showing AI-generated composite risk scores derived from betting pattern analysis, session duration monitoring, chase-loss behaviour indicators, and player-stated self-assessment data with risk tier classification overlays (low, medium, high, at-risk), deposit limit and self-exclusion compliance status display images showing player-set responsible gambling control status, deposit limit breach alert indicators, cooling-off period compliance tracking, and self-exclusion programme registration confirmation displays, PGSI Problem Gambling Severity Index assessment display images showing eight-item gambling consequences and behaviour response scores with total severity category classification (non-problem, low risk, moderate risk, problem gambling), responsible gambling intervention trigger display images showing AI-generated outreach trigger conditions and recommended intervention type (pop-up message, account limit review, support resources referral, mandatory break), and state self-exclusion registry matching status display images showing player identity match result against GAMBLER NJ, iExclusion PA, Nevada Central Registry, and GAMSTOP UK exclusion lists from DraftKings AI at 5.7 million monthly active users processing problem gambling risk score and self-exclusion compliance display images through AI-assisted responsible gambling monitoring and state self-exclusion programme compliance tools; FanDuel AI at 8.5 million daily active users processing PGSI assessment, deposit limit compliance, and intervention trigger display images through AI-assisted responsible gambling programme tools; and Kambi Sports Solutions AI at 50 or more sportsbook operator clients processing responsible gambling monitoring display images through AI-assisted operator responsible gambling programme support tools — extracting NCPG responsible gaming standard compliance determinations, GAMSTOP and state self-exclusion registry match results, PGSI problem gambling severity classifications, and responsible gambling intervention trigger decisions from problem gambling risk score and self-exclusion compliance display image inputs in AI-assisted iGaming responsible gambling programme pipelines.

The adversarial injection surface is the problem gambling risk score display image, PGSI assessment display image, or state self-exclusion registry matching status display image submission pathway: DraftKings AI, FanDuel AI, or Kambi AI responsible gambling monitoring display images submitted through AI-assisted responsible gambling programme compliance and self-exclusion registry matching tools for AI risk tier determination record generation and regulator programme compliance documentation. An adversarially crafted problem gambling risk score display image — in which pixel perturbations applied to the composite risk score numerical display, the risk tier classification colour indicator, the PGSI total score display, the self-exclusion registry match confidence indicator, or the intervention trigger threshold status display cause the AI to classify a player with AI-computed problem gambling indicators meeting the NCPG high-risk or PGSI problem gambling category thresholds — including a player who has set deposit limits and triggered limit breach alerts, who has self-excluded through GAMSTOP or a state self-exclusion registry, or whose betting pattern indicators meet the PGSI “problem gambling” category score of 8 or above — as a low-risk non-problem player not triggering any responsible gambling intervention obligations when the actual betting pattern data and self-exclusion registry matching evidences high-severity problem gambling indicators requiring mandatory intervention — can suppress a responsible gambling alert that would otherwise generate a GAMSTOP match exclusion action, a state self-exclusion programme account suspension, a mandatory PGSI intervention referral, a regulator-required responsible gambling outreach, or a cooling-off period enforcement action. In online gaming operations where DraftKings AI or FanDuel AI processes responsible gambling display images without individual human compliance officer review of every AI-processed risk display before the AI tier determination governs the intervention obligation, adversarial suppression of problem gambling indicators creates NCPG, GAMSTOP, RGSB Safer Gambling Standards, and state self-exclusion programme compliance dimensions.

The NCPG responsible gaming standards, RGSB Responsible Gambling Standards Board Safer Gambling Standards, GAMSTOP UK national self-exclusion registry, and state self-exclusion programme compliance regulatory consequences of adversarially suppressed problem gambling classification span NCPG National Council on Problem Gambling responsible gaming standards and the AGA Responsible Gaming Code of Conduct establishing that US licensed gaming operators must implement problem gambling monitoring, player protection tools, and self-exclusion compliance programmes as conditions of licence compliance in states with responsible gaming statutory requirements — adversarially suppressed problem gambling AI that prevents required interventions creates state gaming licence compliance violation dimensions; RGSB Responsible Gambling Standards Board Safer Gambling Standards (UK) establishing mandatory safer gambling requirements for UKGC-licensed operators including BetMGM UK, with UKGC licence condition requirements for customer interaction and responsible gambling monitoring that carry licence suspension and financial penalty risk — adversarially corrupted UKGC responsible gambling AI creates UKGC Licence Conditions and Codes of Practice (LCCP) compliance violation dimensions with fines that in recent UKGC enforcement actions have reached tens of millions of pounds; GAMSTOP UK national self-exclusion registry matching requirements establishing that all UKGC-licensed online gambling operators must participate in GAMSTOP and must verify customers against the GAMSTOP register at account creation and at regular intervals — adversarially corrupted GAMSTOP matching AI that enables self-excluded customers to continue accessing UK online gaming accounts creates GAMSTOP compliance breach dimensions with UKGC enforcement authority; state self-exclusion programme requirements in NJ (GAMBLER), PA (iExclusion), and Nevada (Central Registry) establishing legal obligations for licensed operators to match customers against state self-exclusion registries and deny access to matched customers — adversarially suppressed state registry matching AI creates state gaming control board licence compliance violation dimensions. Threshold: 50 for responsible gambling problem indicator image injection — reflecting NCPG responsible gaming standards, RGSB Safer Gambling Standards, GAMSTOP UK self-exclusion registry matching, state self-exclusion programme compliance, and PGSI problem gambling severity classification accuracy dimensions.

Integration: iGaming and sports betting AI image ingestion with Glyphward pre-scan

iGaming and sports betting AI image ingestion flows from DraftKings AI, FanDuel AI, and BetMGM AI government ID document scan and KYC display image processing channels, Kambi AI, Sportradar AI, and IGT Advantage AI live odds display dashboard and line movement chart image processing interfaces, DraftKings AI, FanDuel AI, and BetMGM AI AML transaction monitoring alert and SAR documentation display image processing pipelines, and DraftKings AI, FanDuel AI, and Kambi AI problem gambling risk score and self-exclusion compliance status display image processing platforms into KYC age verification AI, odds integrity monitoring AI, AML suspicious activity reporting AI, and responsible gambling programme compliance AI pipelines. Insert Glyphward’s pre-scan at the ingestion boundary before AI-generated output is committed to KYC onboarding eligibility records, odds integrity regulator reports, FinCEN SAR filing determination records, or responsible gambling intervention and self-exclusion programme compliance records:

import asyncio
import base64
import hashlib
import os
import uuid
from enum import Enum
from pathlib import Path

import httpx

GLYPHWARD_API_KEY = os.environ["GLYPHWARD_API_KEY"]
GLYPHWARD_SCAN_URL = "https://glyphward.com/v1/scan"

# iGaming & sports betting AI — adversarial pixel injection in government
# ID KYC document images, odds display dashboards, AML suspicious transaction
# displays, and responsible gambling risk score displays with state Gaming
# Control Board requirements, Wire Act 18 USC §1084, FinCEN BSA 31 USC §5318,
# and GAMSTOP/NCPG responsible gambling standard consequences.

# State Gaming Control Board age verification (NGC Reg 5A, NJ N.J.A.C. 13:69C,
# PA 58 Pa. Code §401a); COPPA 15 USC §6501; OFAC SDN 31 CFR Part 500;
# FinCEN CDD Rule 31 CFR §1010.230 customer due diligence requirements.
THRESHOLD_KYC_AGE_VERIFICATION_AI        = 70

# Wire Act 18 USC §1084 interstate wagering; IGRA 25 USC §2702 tribal gaming
# compact integrity; PASPA-successor state sports betting integrity statutes.
THRESHOLD_ODDS_INTEGRITY_AI              = 55

# FinCEN BSA 31 USC §5318 AML programme; FinCEN CDD 31 CFR §1010.230;
# OFAC SDN 31 CFR Part 500; FinCEN 2021 gaming industry AML guidance.
THRESHOLD_AML_SUSPICIOUS_TRANSACTION_AI  = 65

# NCPG responsible gaming standards; RGSB Safer Gambling Standards (UK);
# GAMSTOP UK national self-exclusion registry; state self-exclusion programmes.
THRESHOLD_RESPONSIBLE_GAMBLING_AI        = 50


class iGamingAIContext(str, Enum):
    KYC_AGE_VERIFICATION_AI        = "kyc_age_verification_ai"        # DraftKings, FanDuel, BetMGM
    ODDS_INTEGRITY_AI              = "odds_integrity_ai"              # Kambi, Sportradar, IGT Advantage
    AML_SUSPICIOUS_TRANSACTION_AI  = "aml_suspicious_transaction_ai"  # DraftKings, FanDuel, BetMGM
    RESPONSIBLE_GAMBLING_AI        = "responsible_gambling_ai"        # DraftKings, FanDuel, Kambi


def threshold_for(context: iGamingAIContext) -> int:
    mapping = {
        iGamingAIContext.KYC_AGE_VERIFICATION_AI:        THRESHOLD_KYC_AGE_VERIFICATION_AI,
        iGamingAIContext.ODDS_INTEGRITY_AI:              THRESHOLD_ODDS_INTEGRITY_AI,
        iGamingAIContext.AML_SUSPICIOUS_TRANSACTION_AI:  THRESHOLD_AML_SUSPICIOUS_TRANSACTION_AI,
        iGamingAIContext.RESPONSIBLE_GAMBLING_AI:        THRESHOLD_RESPONSIBLE_GAMBLING_AI,
    }
    return mapping[context]


async def scan_igaming_ai_image(
    image_path: str | Path,
    context: iGamingAIContext,
    player_entity_hash: str,    # SHA-256 of player account ID or KYC document number
    programme_ref: str,         # e.g. "DRAFTKINGS-KYC-2026-US-NJ-004812", "BETMGM-SAR-2026-0041"
    session_id: str,            # KYC session, odds monitoring session, or AML review batch ID
    client: httpx.AsyncClient,
) -> dict:
    """
    Scan an iGaming or sports betting AI image for adversarial injection
    payloads before forwarding to KYC age verification, odds integrity
    monitoring, AML suspicious activity reporting, or responsible gambling
    programme compliance AI.

    Raises AdversarialiGamingAIImageError if score meets threshold:
      - KYC_AGE_VERIFICATION_AI:        threshold 70; NGC Reg 5A; COPPA 15 USC §6501
      - ODDS_INTEGRITY_AI:              threshold 55; Wire Act 18 USC §1084; IGRA §2702
      - AML_SUSPICIOUS_TRANSACTION_AI:  threshold 65; FinCEN BSA 31 USC §5318
      - RESPONSIBLE_GAMBLING_AI:        threshold 50; NCPG; GAMSTOP UK; state self-exclusion
    """
    image_bytes    = Path(image_path).read_bytes()
    image_b64      = base64.b64encode(image_bytes).decode()
    image_sha256   = hashlib.sha256(image_bytes).hexdigest()
    client_scan_id = str(uuid.uuid4())
    threshold      = threshold_for(context)

    resp = await client.post(
        GLYPHWARD_SCAN_URL,
        headers={"Authorization": f"Bearer {GLYPHWARD_API_KEY}"},
        json={
            "image": image_b64,
            "source": context.value,
            "metadata": {
                "igaming_context":      context.value,
                "player_entity_hash":   player_entity_hash,
                "programme_ref":        programme_ref,
                "session_id":           session_id,
                "client_scan_id":       client_scan_id,
                "image_sha256":         image_sha256,
            },
        },
        timeout=8.0,
    )
    resp.raise_for_status()
    result = resp.json()

    audit_record = {
        "player_entity_hash":  player_entity_hash,
        "programme_ref":       programme_ref,
        "session_id":          session_id,
        "igaming_context":     context.value,
        "scan_id":             result["scan_id"],
        "client_scan_id":      client_scan_id,
        "image_sha256":        image_sha256,
        "score":               result["score"],
        "flagged_region":      result.get("flagged_region"),
        "threshold":           threshold,
        "action":              "blocked" if result["score"] >= threshold else "allowed",
    }
    await write_igaming_audit_record(audit_record)

    if result["score"] >= threshold:
        raise AdversarialiGamingAIImageError(
            f"iGaming AI image blocked [{context.value}]: "
            f"scan_id={result['scan_id']} score={result['score']} "
            f"entity={player_entity_hash} ref={programme_ref}"
        )
    return result


async def write_igaming_audit_record(record: dict) -> None:
    """Persist audit record to iGaming AI regulatory documentation store (stub)."""
    import json, sys
    print(json.dumps(record), file=sys.stderr)


class AdversarialiGamingAIImageError(Exception):
    """Raised when an iGaming or sports betting AI image exceeds the adversarial injection threshold."""
    pass

Call scan_igaming_ai_image() with iGamingAIContext.KYC_AGE_VERIFICATION_AI before forwarding DraftKings AI, FanDuel AI, or BetMGM AI government ID document scan and facial comparison display images to KYC age verification and onboarding AI — with player_entity_hash as the SHA-256 of the player account ID and programme_ref as the KYC session identifier for state Gaming Control Board age verification compliance, COPPA 15 USC §6501 minor protection, OFAC SDN 31 CFR Part 500 sanctions screening, and FinCEN CDD 31 CFR §1010.230 customer due diligence audit trail. Call with iGamingAIContext.ODDS_INTEGRITY_AI for Kambi AI, Sportradar AI, or IGT Advantage AI live odds display dashboard and line movement chart images before odds integrity monitoring AI — for Wire Act 18 USC §1084 interstate wagering compliance, IGRA 25 USC §2702 tribal gaming compact integrity, and state sports betting statute integrity reporting audit trail. Call with iGamingAIContext.AML_SUSPICIOUS_TRANSACTION_AI for transaction monitoring alert and SAR display images before AML reporting AI — for FinCEN BSA 31 USC §5318 SAR filing obligation compliance, OFAC SDN screening audit trail, and FinCEN 2021 gaming AML guidance documentation. Call with iGamingAIContext.RESPONSIBLE_GAMBLING_AI for problem gambling risk score, PGSI assessment, and self-exclusion status display images before responsible gambling programme AI — for NCPG responsible gaming standard compliance, GAMSTOP UK national self-exclusion registry matching accuracy, and state self-exclusion programme compliance. Get early access

Coverage matrix

Related questions

Further reading

• FigStep adversarial image injection detection — technical overview of pixel-level adversarial perturbation underlying government ID KYC document injection, odds display dashboard manipulation, and AML transaction monitoring alert suppression in iGaming AI.
• Vision-language model security — architectural overview of multimodal AI adversarial injection covering the VLM image encoder layers that DraftKings AI, FanDuel AI, Kambi AI, and Sportradar AI use to process identity verification, odds display, and compliance indicator images.
• Free tier — 10 scans/day, no card required — start scanning iGaming and sports betting AI image inputs at development volumes; test KYC document, odds integrity dashboard, AML SAR display, and PGSI responsible gambling injection detection without a payment method on file.
• Prompt injection in sports analytics AI — related sports AI injection surface covering sports data and analytics AI with overlapping Sportradar AI integrity monitoring and sports betting market integrity dimensions.
• Prompt injection in fintech and payments AI — related payments and AML AI injection surface covering FinCEN BSA, OFAC sanctions screening, and payment fraud detection AI with overlapping iGaming AML transaction monitoring dimensions.
• Prompt injection in cryptocurrency and blockchain AI — related financial compliance AI injection surface covering crypto AML and OFAC screening AI with overlapping FinCEN BSA 31 USC §5318 and SDN list matching dimensions.
• Prompt injection in employee background screening AI — related identity verification AI injection surface covering background check document analysis AI with overlapping government ID document scan, facial comparison, and identity authenticity assessment dimensions.