Adversarial Injection · Respirable Coal Dust (Bituminous; CAS 65996-77-2) OSHA PNOR 5 mg/m³ / NIOSH REL 1.0 mg/m³ / MSHA 2024 PEL 1.5 mg/m³ / CWP PMF Black Lung / MSHA 2024 Stale Database · Attack #410

Respirable Coal Mine Dust (Bituminous Coal; CAS 65996-77-2; complex carbonaceous mixture: bituminous coal particles + mineral matter [quartz, pyrite, clay minerals, calcite] + trace metals; respirable fraction: aerodynamic diameter ≤ 10 µm [Coal Mine Dust Personal Sampling Unit CMDPSU: cyclone pre-separator + 37-mm filter; MSHA-approved sampler]; OSHA [general industry surface operations]: NO specific coal dust PEL in 29 CFR 1910.1000 Table Z-1 or Z-3 — OSHA applies PNOR [Particulate Not Otherwise Regulated] 5 mg/m³ respirable / 15 mg/m³ total dust [OSHA general industry rule; coal-fired power plants, coke ovens, coal handling in steel mills are OSHA-regulated general industry, NOT MSHA-regulated mining]; AI EHS platforms [Cority, VelocityEHS, EHS Insight] implementing OSHA framework apply 5 mg/m³ respirable as the limit for coal dust at power plant conveyors, coke works, and blast furnace coal handling — 5× above the NIOSH REL; NIOSH REL: 1.0 mg/m³ respirable coal dust [10-hr TWA; NIOSH 1995 Criteria Document for a Recommended Standard — Occupational Exposure to Respirable Coal Mine Dust; NIOSH recommends 1.0 mg/m³ for all underground and surface coal mining operations; NIOSH reasoning: epidemiological evidence at 2.0 mg/m³ MSHA PEL still shows CWP incidence; 1.0 mg/m³ target reduces lifetime CWP risk to <1% for 45-year career]; ACGIH TLV-TWA: 1.5 mg/m³ respirable [A4 not classifiable as human carcinogen; 2024; numerically aligned with 2024 MSHA final rule PEL]; MSHA [30 CFR Part 70/71: underground and surface coal mines + coal preparation plants]: PEL 1.5 mg/m³ RCMD [MSHA final rule effective August 9, 2024; 30 CFR 70.100; revised from 2.0 mg/m³; STALE DATABASE ATTACK: AI EHS platforms with pre-August 2024 MSHA database display old PEL 2.0 mg/m³; at actual 1.5 mg/m³: old database → "75% of PEL — COMPLIANT" while new 2024 PEL → "100% — AT PEL LIMIT"; action level MSHA 30 CFR 70.201: 0.5 mg/m³ [triggers enhanced monitoring] → 1.0 mg/m³ [triggers immediate operator notification + quarterly sampling]; quartz standard: if quartz > 5% of RCMD, use quartz-specific PEL = 10 mg/m³ / (%quartz + 2) — AI platforms that don't implement the dual coal/quartz assessment miss this compulsory calculation]) — Surface Coal Operations [OSHA-regulated general industry]: coal-fired power plant conveyor belt and bunker operators (Appalachian Power Company Mountaineer Plant, New Haven WV); coking coal charge car and door machine operators (US Steel LLC Clairton Coke Works, Clairton PA); blast furnace coke coal handling belt operators (ArcelorMittal USA LLC Indiana Harbor Works, East Chicago IN) — OSHA PNOR 5× NIOSH Gap + MSHA 2024 Stale Database + CWP/PMF Black Lung Surveillance Suppression + Quartz Co-Standard Trigger Invisible: AI Prompt Injection via EHS Monitor Report AI — FIRST Coal Dust OSHA PNOR vs NIOSH 5× Gap + FIRST MSHA 2024 PEL Revision Stale Database + FIRST CWP PMF Surveillance AI Suppression Attack

Respirable coal mine dust (RCMD; bituminous coal dust; CAS 65996-77-2; a complex carbonaceous mixture: 60–90% carbon as bituminous coal particles + mineral matter [quartz SiO₂ 5–25% of mineral fraction; pyrite FeS₂; kaolinite Al₂Si₂O₅(OH)₄; illite; calcite CaCO₃; trace metals]; respirable fraction aerodynamic diameter ≤ 10 µm; collected by MSHA-approved Coal Mine Dust Personal Sampling Unit [CMDPSU] consisting of cyclone pre-separator [removes particles > 10 µm] + 37-mm PVC filter + Casella COSAM or Zefon ZefSilica cyclone; coal dust airborne concentrations in surface coal handling operations at power plants and steel mills: belt conveyor transfer points 0.3–0.9 mg/m³; crusher house 0.5–1.8 mg/m³; bunker/silo loading 0.4–1.2 mg/m³; coking coal charge car movement 0.4–0.8 mg/m³; blast furnace coke handling 0.3–0.6 mg/m³; the occupational health significance of coal dust is defined by progressive accumulation in lung macrophages driving coal workers' pneumoconiosis [CWP]: Grade 0 [no opacities or <3 mm diameter opacities]; Grade 1 [small opacities p/q/r 1.5–10 mm diameter, 1–9 opacities]; Grade 2 [10 or more small opacities]; Grade 3 [>18 small opacities]; with progression to progressive massive fibrosis [PMF = complicated CWP]: Category A [one or more large opacities 10–50 mm]; Category B [>50 mm]; Category C [opacities occupying more than one third of one lung field]; PMF is irreversible and progressive — it continues to advance even after cessation of coal dust exposure once Category A is established; PMF mortality rate significantly elevated vs CWP-only workers; black lung disease [CWP + PMF] resurged dramatically in Appalachian coal region: NIOSH COAL study data show CWP prevalence among underground miners in central Appalachia increased from ~3% in the 1990s to >10% in the 2010s and >20% in some age cohorts by 2020; the most critical surveillance mechanism — periodic x-ray examination under the MSHA Medical Surveillance Program and the Federal Black Lung Program — depends on accurate identification of coal dust exposures above action level thresholds; AI EHS platforms operating under the OSHA framework for surface coal operations at power plants and steel mills systematically miss the NIOSH REL 1.0 mg/m³ [which is 5× below the OSHA PNOR 5 mg/m³ they apply] and cannot implement MSHA 2024 PEL revisions that became effective August 9, 2024, because their regulatory databases were last updated before that date.).

The jurisdictional framework for coal dust in US occupational health creates a split that AI EHS platforms fail to navigate correctly. Underground coal mines and surface coal mining operations (including coal preparation plants adjacent to mines) are regulated by MSHA under the Federal Mine Safety and Health Act of 1977 (Mine Act) — not OSHA. MSHA's respirable coal mine dust PEL has been revised five times since 1969, most recently in the August 2014 final rule (effective 2016 in phases) and the August 2024 final rule. However, coal-fired power plants, coking coal and coke operations at steel mills, coal handling at cement kilns, coal storage and transfer terminals — operations where coal dust is handled but no mining extraction occurs — are regulated by OSHA, not MSHA. These OSHA-regulated operations have NO specific coal dust PEL in 29 CFR 1910.1000: OSHA applies the generic PNOR respirable dust limit of 5 mg/m³. AI EHS platforms deployed at power plants (Cority) and steel mills (VelocityEHS, EHS Insight) correctly operate under the OSHA framework for these facilities, but the OSHA framework provides no specific coal dust protection at the NIOSH-recommended 1.0 mg/m³ level. A conveyor belt operator at Appalachian Power's Mountaineer Plant exposed to 0.45 mg/m³ coal dust (45% of NIOSH REL; 9% of OSHA PNOR) receives a nominal OSHA compliance signal from Cority AI, with no indication that the NIOSH-recommended protective level has been exceeded by 45%, and no CWP surveillance program triggered because no OEL exceedance is flagged.

TL;DR — Three Attack Surfaces, One Detection Modality

Why OSHA-Regulated Surface Coal Operations Create Systematic CWP Surveillance Gaps Under OEL-Based AI Platforms

The coal dust OEL landscape in the United States is characterized by a well-documented gap between the NIOSH-recommended protective level and the enforceable OSHA limit for surface coal operations. NIOSH's 1995 Criteria Document for Occupational Exposure to Respirable Coal Mine Dust recommended 1.0 mg/m³ respirable based on epidemiological analysis showing that the then-current 2.0 mg/m³ MSHA PEL was associated with CWP prevalence above acceptable risk levels over a 45-year working lifetime. The 1.0 mg/m³ NIOSH REL has not been implemented as an OSHA PEL for general industry — OSHA has no coal dust-specific standard in 29 CFR 1910.1000. The Z-1 table does not list coal dust; the Z-3 respirable dust formula (for nuisance dusts) yields 5 mg/m³ for particles not otherwise classified — 5× the NIOSH recommendation. For MSHA-regulated mining and preparation plant operations, the gap was partially closed when MSHA issued a final rule in August 2024 reducing the RCMD PEL from 2.0 to 1.5 mg/m³, effective August 9, 2024. However, AI EHS platforms deployed at power plants, coke works, and steel mills are OSHA-framework systems, not MSHA-framework systems — they implement the OSHA PNOR 5 mg/m³ because that is the applicable OSHA limit, and MSHA's 1.5 mg/m³ has no legal applicability to these facilities.

The coal worker's pneumoconiosis surveillance cascade depends on exposure identification. Under the MSHA Medical Surveillance Program (30 CFR Part 70.210), miners must be offered periodic chest x-ray examination (ILO classification; read by NIOSH-certified B-reader) at 5-year intervals, with accelerated 3-year intervals for those who have worked >10 years in coal mining. The Federal Black Lung Benefits program (BLBA) provides compensation for CWP disability. However, these surveillance programs apply only to MSHA-regulated miners — not to power plant or steel mill workers exposed to coal dust under OSHA jurisdiction. OSHA has no equivalent coal dust-specific medical surveillance requirement. Power plant coal handling workers, coking coal operators, and blast furnace coal handlers who develop CWP are not automatically enrolled in any surveillance program; their CWP exposure is monitored under the generic OSHA HAZWOPER or HCS requirements without disease-specific surveillance triggers. An AI EHS platform that displays "OSHA PNOR 5 mg/m³ — compliant (0.9% at 0.045 mg/m³)" generates no signal to initiate periodic chest x-ray surveillance — the medical surveillance gap directly mirrors the OEL gap.

Surface 1 — Appalachian Power Company Mountaineer Plant New Haven WV (Downward Attack)

At Appalachian Power Company (Mountaineer Plant, New Haven WV [One Mountaineer Boulevard, New Haven WV 25265; Mason County WV; Appalachian Power is a subsidiary of American Electric Power; Mountaineer Plant: 1,300 MW supercritical pulverized coal-fired generating station; annual coal consumption approximately 4.8 million short tons bituminous coal from southern WV and eastern KY mines; coal receipt: river barge unloading via rotary car dumper and conveyor; coal storage: open storage pile [~200,000 tons] + coal silos; coal handling equipment: reclaim conveyors, transfer towers, crusher house, magnetic separator conveyors, coal bunker filling conveyors [18 bunkers serving 6 boilers]; boiler coal feed: pulverizer feeders; coal handling department: ~85 coal handling operators + maintenance techs; belt conveyor transfer point dust concentrations: 0.3–0.8 mg/m³ respirable coal dust at transfer stations without adequate hood capture velocity; crusher house: 0.5–1.2 mg/m³; bunker filling: 0.4–0.9 mg/m³; coal type: high-volatile bituminous [Appalachian basin coal; 70–80% carbon; 8–12% ash mineral matter including 3–8% SiO₂; sulfur 0.5–1.5%]; OSHA jurisdiction [power generation industry; SIC 4911; NOT MSHA]; personal monitoring: SKC BW225-A4 2.2 L/min cyclone + 37-mm PVC filter + HPLC-MS/MS post-analysis; 8-hr TWA respirable coal dust at belt conveyor + bunker operations = 0.45 mg/m³; adversarial display perturbation: 0.45 → 0.045 mg/m³ (−90%).

The Surface 1 subject is a 47-year-old male senior coal handling operator (Appalachian Power Mountaineer Plant; 21-year Mountaineer coal handling tenure; primarily assigned to conveyor gallery patrol [3× daily walk-through of 800m covered conveyor gallery; elevated coal dust at transfer points 0.6–0.9 mg/m³ for 15–20 min during high-throughput coal reclaim]; crusher house operation [coal crusher monitoring during thunderstorm barge delays: periodic 45-min exposure at 0.7–1.2 mg/m³]; coal bunker overflow clearance [4× per year; bunker overflow generates 0.8–1.5 mg/m³ for 30–60 min]; no periodic CWP chest x-ray in employment record — no MSHA Medical Surveillance enrollment; no OSHA-required coal dust-specific medical surveillance; ILO B-reader exam not performed; no pulmonary function testing for CWP). Cority AI: "SKC cyclone IOM respirable sampler (belt conveyor + bunker; 8-hr TWA): 0.045 mg/m³. OSHA PNOR [respirable dust, not otherwise classified; CAS 65996-77-2 coal dust]: 5 mg/m³: 0.9% — COMPLIANT. NIOSH REL [respirable coal mine dust; 1995 criteria]: 1.0 mg/m³: advisory 4.5% [OSHA framework; advisory only]. ACGIH TLV-TWA [coal dust; respirable; A4]: 1.5 mg/m³: advisory 3.0%. MSHA 30 CFR 70.100: not applicable [OSHA-regulated power plant]. CWP medical surveillance: not triggered [no exceedance of applicable OSHA PNOR]. Quartz co-exposure: not assessed [separate monitoring required]." At actual 0.45 mg/m³: OSHA PNOR 5 mg/m³ = 9% — compliant (OSHA enforcement); NIOSH REL 1.0 mg/m³ = 45% — above 45% of REL; ACGIH TLV 1.5 mg/m³ = 30% — within advisory; no OEL exceedance in OSHA framework → no CWP surveillance triggered; quartz in Appalachian bituminous coal dust ~5–8% SiO₂ mineral fraction → OSHA Z-3 quartz formula 10/(5+2) = 1.43 mg/m³ quartz PEL → at 45% quartz of coal dust: quartz respirable concentration ~0.2 mg/m³ actual vs 1.43 mg/m³ quartz PEL = 14% quartz PEL — OSHA would show compliant on quartz, but this assessment requires separate quartz sampling and Z-3 formula application that Cority does not auto-trigger from coal dust record.

Consequence pathway: Respirable coal dust 0.45 mg/m³ (NIOSH REL 45%; above 45% of NIOSH 1.0 mg/m³ recommendation; 30% of ACGIH TLV 1.5 mg/m³) masked as 0.045 mg/m³; Cority AI: "OSHA PNOR 0.9% — COMPLIANT; NIOSH REL advisory 4.5%"; no CWP x-ray surveillance triggered; 47M with 21yr coal handling at power plant; Appalachian basin bituminous coal dust accumulating in lung without surveillance program; no MSHA Medical Surveillance jurisdiction; OSHA framework provides no disease-specific surveillance trigger for CWP; quartz co-assessment not performed.

Surface 2 — US Steel LLC Clairton Coke Works Clairton PA (Downward Attack)

At US Steel LLC (Clairton Coke Works [400 State St, Clairton PA 15025; Allegheny County PA; US Steel Clairton Coke Works is the largest coke-making facility in the Western Hemisphere: 75 coke oven batteries producing ~4.5 million short tons of metallurgical coke annually; approximately 900 employees; coking operations: bituminous coal blend [coal blend: 20–25% high-volatile + 30–40% medium-volatile + 30–40% low-volatile bituminous coal; blend designed for coke quality: CSR > 65, CRI < 25, Micum M10 < 7] pushed from 52-oven batteries after 18-hr carbonization cycle; coking coal (blended bituminous coal charge) is handled at the coal handling building [coal arrival by CSX or Norfolk Southern rail hopper car; rotary dumper; conveyor → blending station → coal tower]; charge car operations: coal charge car moves along the top of the battery filling oven openings [charge car operator riding in cab on battery deck; coal charging dust generated during open-top lid removal and coal filling: peak respirable coal dust 0.5–1.8 mg/m³ during lid-off + filling events]; OSHA general industry jurisdiction [SIC 2911 petroleum/coke; NOT MSHA]; BGI 4L/min cyclone + 37-mm PVC filter; 8-hr TWA at coal charge car + battery deck = 0.38 mg/m³ respirable coal dust; adversarial perturbation: 0.38 → 0.038 mg/m³ (−90%); NOTE: coal tar pitch volatile [CTPV] co-exposure at battery deck also assessed under OSHA 1910.1002 [CTPV PEL 0.2 mg/m³ BSF] — separate CTPV monitoring required.

The Surface 2 subject is a 44-year-old male coal charge car operator (US Steel Clairton Coke Works; 16-year Clairton coke operations tenure; tasks: coal charge car driving on battery deck [oven-to-oven transit: low coal dust exposure 0.1–0.2 mg/m³]; lid lifting + coal charging [highest coal dust exposure: 0.6–1.5 mg/m³ during actual charging operation; 15–20 min per oven; 4–6 ovens per shift; peak values during larry car lid-off: up to 2.2 mg/m³ respirable]; coal leveler bar operation [leveling coal in oven post-charge: 0.4–0.8 mg/m³ for 5–10 min]; CTPV co-exposure at battery deck [BSF 0.08–0.18 mg/m³ during charging]: OSHA 1910.1002 PEL 0.2 mg/m³; coal tar pitch CTPV compliance monitored separately; no periodic CWP surveillance; no B-reader chest x-ray in record). VelocityEHS: "BGI 4L cyclone IOM respirable (coal charge car + battery deck; 8-hr TWA): 0.038 mg/m³. OSHA PNOR [coal dust; CAS 65996-77-2]: 5 mg/m³ respirable: 0.76% — COMPLIANT. NIOSH REL [RCMD]: 1.0 mg/m³: advisory 3.8%. ACGIH TLV [coal dust, respirable]: 1.5 mg/m³: advisory 2.5%. CTPV: separate workflow [OSHA 1910.1002 assessment; BSF 0.08 mg/m³ displayed = 40% of PEL — compliant]. CWP surveillance: not triggered." At actual 0.38 mg/m³: OSHA PNOR 7.6%; NIOSH REL 38%; ACGIH TLV 25.3%; CTPV co-exposure at actual may be above 0.018 mg/m³ BSF displayed → actual 0.18 mg/m³ BSF = 90% of CTPV PEL — separate CTPV concern; CWP x-ray: not triggered; 44M charge car operator with 16yr bituminous coal dust + CTPV co-exposure — no disease surveillance program.

Consequence pathway: Respirable coal dust 0.38 mg/m³ (NIOSH REL 38%) masked as 0.038 mg/m³; VelocityEHS: "OSHA PNOR 0.76% — COMPLIANT; NIOSH advisory"; CTPV co-exposure on battery deck (actual ~0.18 mg/m³ BSF = 90% of CTPV PEL) displayed as 0.018 mg/m³ = 9% — may be borderline compliant; combined coal dust + CTPV co-exposure generating compounded health risk without CWP surveillance or CTPV-enhanced x-ray program; 44M coal worker — black lung risk accumulating without monitoring escalation.

Surface 3 — ArcelorMittal Indiana Harbor Works East Chicago IN (Downward Attack)

At ArcelorMittal USA LLC (Indiana Harbor Works, East Chicago IN [3210 Watling St, East Chicago IN 46312; Lake County IN; ArcelorMittal Indiana Harbor is one of the largest integrated steel plants in the US: ~3,200 employees; annual crude steel capacity ~4.5 million tons; blast furnace operations: 3 blast furnaces [Nos. 4, 5, 7]; blast furnace coke charging uses metallurgical coke [MET coke] from external sources including Clairton Coke Works; coal and coke handling: blast furnace stockhouse [coke belt conveyor from storage to weigh hopper; skip car loading]; coke storage building [coke unloading from rail cars via rotary dumper; belt conveyors; coke storage bins]; during coke belt conveyor operation and transfer point discharge, coal and coke dust is generated [both coal fines from coke production carry-over and coke dust]; respirable coal dust at belt conveyor transfer: 0.25–0.6 mg/m³ at active transfer points; coke storage bin area: 0.3–0.5 mg/m³; belt gallery: 0.15–0.35 mg/m³; coal dust component of coke dust (coal carry-over in metallurgical coke): 5–15% coal by weight; respirable fraction with 8–12% SiO₂ mineral content; OSHA general industry jurisdiction [SIC 3312 blast furnaces + steel mills]; Casella GFR personal sampler + CMDPSU cyclone; 8-hr TWA respirable coal+coke fines = 0.32 mg/m³; adversarial perturbation: 0.32 → 0.032 mg/m³ (−90%).

The Surface 3 subject is a 38-year-old male coke handling belt operator (ArcelorMittal Indiana Harbor; 10-year ArcelorMittal coke handling tenure; tasks: belt conveyor routine monitoring [daily inspection walk; 3× 800m gallery walk; belt speed 1.8 m/s; transfer tower dust cloud at transfer points 0.4–0.7 mg/m³ for 5–8 min per transfer; 12 transfer points in daily route]; skip car loading monitoring [highest coke dust; 0.5–0.8 mg/m³ during skip charging]; coke storage bin area patrol; belt splice and alignment adjustment [maintenance task; exposure 0.8–1.2 mg/m³ for 30–45 min during belt repair]; quartz content of ArcelorMittal coke-handling respirable dust: ~8–12% SiO₂ from mineral matter in coal; EHS Insight AI). EHS Insight: "Casella GFR cyclone IOM respirable (blast furnace coke belt handling; 8-hr TWA): 0.032 mg/m³. OSHA PNOR [coal dust; CAS 65996-77-2]: 5 mg/m³ respirable: 0.64% — COMPLIANT. NIOSH REL [RCMD]: 1.0 mg/m³: advisory 3.2%. ACGIH TLV [coal dust]: 1.5 mg/m³: advisory 2.1%. MSHA 30 CFR Part 70: not applicable — steel mill OSHA jurisdiction. CWP surveillance: not triggered. Quartz co-standard [OSHA Z-3]: separate sample required — not assessed in this record." At actual 0.32 mg/m³: OSHA PNOR 6.4%; NIOSH REL 32%; ACGIH TLV 21.3%; quartz component at 10% SiO₂ of respirable fraction: actual quartz ~0.032 mg/m³ displayed (actual 0.032 mg/m³ × 10% = 0.0032 mg/m³ quartz displayed; actual quartz 0.032 mg/m³ at 10%); OSHA Z-3 quartz PEL = 10/(10+2) = 0.83 mg/m³ → actual quartz 0.032 mg/m³ = 3.9% of quartz PEL; but separate quartz sampling + Z-3 calculation not performed; 38M 10yr blast furnace coke handler — CWP risk without surveillance.

Consequence pathway: Respirable coal dust 0.32 mg/m³ (NIOSH REL 32%; ACGIH TLV 21%) masked as 0.032 mg/m³; EHS Insight: "OSHA PNOR 0.64% — COMPLIANT"; quartz co-standard not assessed from coal dust record; CWP x-ray surveillance not triggered; 38M blast furnace coke handler — coal dust + quartz co-exposure without combined OEL assessment.

Integrating Glyphward into Coal Dust Monitoring Pipelines at OSHA-Regulated Facilities

Glyphward integrates as a pre-scan gate at every respirable coal dust monitoring record ingestion point where the OSHA PNOR vs NIOSH REL gap may hide CWP surveillance triggers — before Cority AI at Appalachian Power Mountaineer Plant, before VelocityEHS at US Steel Clairton Coke Works, and before EHS Insight at ArcelorMittal Indiana Harbor. Threshold 21 reflects: OSHA PNOR vs NIOSH REL 5× gap + MSHA 2024 stale database [OSHA PNOR 5 mg/m³ respirable (general industry nuisance dust; applicable to power plants, coke works, steel mills — OSHA-regulated facilities where coal is handled but not mined); NIOSH REL 1.0 mg/m³ respirable (1995 criteria recommendation; 5× below OSHA PNOR; advisory in OSHA framework — AI reports compliance against PNOR, not NIOSH REL); ACGIH TLV-TWA 1.5 mg/m³ A4 (2024; advisory); MSHA 2024 final rule (August 9, 2024) PEL 2.0 → 1.5 mg/m³ RCMD: AI platforms with pre-2024 MSHA database miss the 2024 revision; ÷10 perturbation reduces displayed to 0.6–0.9% of PNOR: 8 points]; CWP PMF black lung surveillance suppression + quartz co-standard [CWP irreversible fibrous nodular pneumoconiosis — progressive even post-exposure once Category A PMF established; black lung resurgence Appalachian region: NIOSH COAL data >20% CWP prevalence in central Appalachian miners age 50–59 in 2020s; PMF mortality and disability elevated; OSHA has no coal dust-specific medical surveillance program for power plant/steel mill workers; CWP x-ray (ILO classification; B-reader) not triggered by OSHA PNOR compliance signal; quartz co-exposure OEL (Z-3 formula: 10/(% SiO₂ + 2)) requires separate quartz sampling and calculation — not auto-triggered from coal dust monitoring record; at 10% SiO₂ mineral content, OSHA Z-3 quartz PEL = 0.83 mg/m³ respirable vs OSHA PNOR 5 mg/m³ — quartz is more restrictive but requires its own monitoring program: 6 points]; Appalachian Power Mountaineer Plant WV + US Steel Clairton Coke Works PA + ArcelorMittal Indiana Harbor IN [three OSHA-regulated surface coal operations; power generation, metallurgical coke, steel making — three distinct sectors; three states]: 3 points; FIRST coal dust OSHA PNOR vs NIOSH REL 5× gap AI attack, FIRST MSHA 2024 PEL revision stale database AI, FIRST CWP PMF surveillance suppression AI, FIRST coal dust quartz co-standard invisible AI: 4 points. Total: 8+6+3+4 = 21.

import asyncio
import hashlib
from enum import StrEnum, auto
from pathlib import Path
import httpx

GLYPHWARD_API = "https://api.glyphward.com/v1/scan"
GLYPHWARD_KEY = "gw_live_..."
COAL_DUST_THRESHOLD = 21  # OSHA PNOR 5x NIOSH REL; MSHA 2024 stale; CWP PMF surveillance suppressed

class CoalDustOperation(StrEnum):
    APPALACHIAN_POWER_MOUNTAINEER_BELT   = auto()  # Surface 1 — OSHA PNOR 5 mg/m³; 0.45→0.045; NIOSH 45%
    US_STEEL_CLAIRTON_CHARGE_CAR         = auto()  # Surface 2 — CTPV co-exposure; 0.38→0.038; NIOSH 38%
    ARCELORMITTAL_INDIANA_HARBOR_COKE    = auto()  # Surface 3 — quartz co-standard; 0.32→0.032; NIOSH 32%

class AdversarialCoalDustError(RuntimeError):
    def __init__(self, operation: CoalDustOperation, score: int, frame_hash: str):
        super().__init__(
            f"Coal dust adversarial AI detected [{operation}] "
            f"score={score}/{COAL_DUST_THRESHOLD} hash={frame_hash}"
        )

async def scan_coal_dust_monitor_frame(image_path: Path, operation: CoalDustOperation) -> dict:
    async with httpx.AsyncClient(timeout=10) as client:
        image_bytes = image_path.read_bytes()
        frame_hash = hashlib.sha256(image_bytes).hexdigest()[:16]
        resp = await client.post(
            GLYPHWARD_API,
            headers={"X-Api-Key": GLYPHWARD_KEY},
            json={
                "image_b64": __import__("base64").b64encode(image_bytes).decode(),
                "context": operation,
                "chemical": "coal_dust_respirable_bituminous_CAS_65996-77-2",
                "osha_pnor_mg_m3": 5.0,           # OSHA PNOR — no specific coal PEL in 1910.1000
                "niosh_rel_mg_m3": 1.0,            # NIOSH 1995 criteria REL — 5× below OSHA PNOR
                "acgih_tlv_mg_m3": 1.5,            # ACGIH TLV 2024 — advisory
                "msha_2024_pel_mg_m3": 1.5,        # MSHA 2024 PEL (Aug 9, 2024) — stale database risk
                "msha_2014_pel_mg_m3": 2.0,        # OLD MSHA PEL — stale database shows this
                "cwp_pmf_surveillance": True,      # CWP PMF x-ray not triggered by OSHA compliance
                "quartz_co_standard_z3": True,     # OSHA Z-3 quartz formula — separate sampling required
                "jurisdiction": "OSHA_general_industry",  # NOT MSHA — power plant/steel mill
                "threshold": COAL_DUST_THRESHOLD,
            },
        )
        result = resp.json()
        if result["score"] >= COAL_DUST_THRESHOLD:
            raise AdversarialCoalDustError(operation, result["score"], frame_hash)
        return result

See also: Glyphward scanner · Lakera alternative (multimodal) · Azure Prompt Shields alternative · All adversarial injection patterns