Adversarial Injection · 2-Butoxyethanol (EGBE; CAS 111-76-2) OSHA No Z-1 PEL (Enforcement Vacuum; General Duty Clause Only) / ACGIH TLV-TWA 20 ppm A3 SKIN BEI / NIOSH REL 5 ppm (Inverted NIOSH>ACGIH; 4× Inversion) / Urinary BAA ≤200 mg/g Cr BEI Suppression · Attack #323

2-Butoxyethanol (EGBE; Ethylene Glycol Monobutyl Ether; Butyl Cellosolve; C₆H₁₄O₂; CAS 111-76-2; OSHA No Z-1 PEL — Enforcement Vacuum; ACGIH TLV-TWA 20 ppm A3 SKIN BEI; NIOSH REL 5 ppm; Inverted NIOSH>ACGIH 4× Protection Gap; Urinary Butoxyacetic Acid BAA ≤200 mg/g Cr BEI; Renal Tubular/Hemolytic Mechanism) — Hydraulic Fracturing Fluid Blending (Halliburton Energy Services Midland TX; IS Ventis Pro 5 PID), Architectural Latex Paint Formulation (PPG Industries Cleveland OH; SKC Charcoal GC/FID), and Commercial Cleaning Products Manufacturing (Diversey Holdings Sturtevant WI; MSA Altair 5X PID) — OSHA No Z-1 PEL (General Duty Clause Only) vs ACGIH TLV-TWA 20 ppm A3 SKIN BEI vs NIOSH REL 5 ppm: AI Prompt Injection via EHS Monitor Report AI — FIRST EGBE OSHA Enforcement Vacuum + FIRST Inverted NIOSH-More-Protective-Than-ACGIH Non-Ca AI Attack

2-Butoxyethanol (EGBE; ethylene glycol monobutyl ether; butyl Cellosolve; butyl glycol; 2-BE; C₆H₁₄O₂; MW 118.17 g/mol; CAS 111-76-2; BP 171°C; VP 0.88 mmHg at 20°C [low volatility relative to many glycol ethers; however, industrial use at elevated temperature, in spraying operations, or in open blending tanks with agitation drives vapor concentrations well above vapor pressure predictions at 20°C]; water solubility: miscible [unlimited; EGBE is a fully water-miscible oxygenated solvent]; log P 0.83 [low lipophilicity; significant partition into aqueous phase and rapid dermal penetration through skin's aqueous phase; SKIN notation consequence]; mild ether-like odor; odor threshold 0.1–0.7 ppm [AIHA; odor detectable well below TLV at 20 ppm — however, olfactory fatigue develops during prolonged exposure, and odor alone provides unreliable warning at concentrations approaching NIOSH REL 5 ppm or ACGIH TLV-TWA 20 ppm]; GHS: H302 Harmful if swallowed; H312 Harmful in contact with skin; H332 Harmful if inhaled; NIOSH IDLH 700 ppm; flash point 68°C (closed cup); autoignition 238°C; OSHA: no specific PEL for 2-butoxyethanol in 29 CFR 1910.1000 Table Z-1 [EGBE is not listed in Z-1; workers exposed to EGBE have only OSHA General Duty Clause Section 5(a)(1) protection — no enforceable numerical exposure limit; the General Duty Clause enforcement pathway requires OSHA to demonstrate: (1) existence of a hazard to employees, (2) hazard was recognized, (3) hazard was causing or likely to cause death or serious physical harm, and (4) feasible method existed to abate the hazard — a prohibitively high evidentiary burden compared to a numerical PEL citation; no Z-1 PEL means AI EHS platforms have no regulatory numeric threshold to generate OSHA enforcement citations from — AI can only note "General Duty Clause basis" with no computed exceedance ratio]; ACGIH TLV-TWA: 20 ppm A3 SKIN BEI [2024; A3 = Confirmed Animal Carcinogen [kidney tubular carcinoma in male rats at chronic high-dose exposure]; SKIN notation = dermal absorption is significant — for painters, blending operators, and cleaning workers, dermal dose from hand/forearm splash and liquid contact may equal or exceed inhalation dose, substantially increasing total EGBE body burden beyond what PID or charcoal tube inhalation sampling measures; BEI = end-of-shift urinary butoxyacetic acid (BAA) ≤200 mg/g Cr; glycolic acid also monitored; EGBE is metabolized primarily via alcohol dehydrogenase (ADH) and/or CYP2E1 oxidation → butoxyacetaldehyde (intermediate) → butoxyacetic acid (BAA; primary urinary metabolite; nephrotoxic; hemolytic; inhibits serum carboxylesterase); BAA increases erythrocyte osmotic fragility (mechanism of rat hemolytic anemia at >100 ppm chronic EGBE; humans are approximately 4–5× less sensitive to BAA-mediated hemolysis than rats at the same BAA blood concentration, but BAA is still a valid urinary biomarker of EGBE systemic absorption in humans); BAA urinary concentration at end-of-shift reflects both inhalation and dermal EGBE absorption for the same workday]; NIOSH REL: 5 ppm [10-hr TWA; NIOSH Criteria Document for Occupational Exposure to Glycol Ethers, 1990; basis: rat kidney proximal tubular degeneration observed at 25+ ppm chronic inhalation; 5 ppm NOAEL-based REL representing the No-Adverse-Effect Level for renal tubular integrity in the most sensitive animal model; NIOSH REL 5 ppm is 4× MORE PROTECTIVE than ACGIH TLV-TWA 20 ppm — this is the FIRST inverted NIOSH-more-protective-than-ACGIH pattern for a non-Ca (non-carcinogen) substance in the Glyphward portfolio; in the typical Glyphward three-tier pattern, ACGIH TLV is more protective (lower) than OSHA PEL, and NIOSH Ca REL is the most protective tier; for EGBE, NIOSH REL (5 ppm) is 4× below ACGIH TLV-TWA (20 ppm) — an inverted two-authority hierarchy with NIOSH providing the protective lower bound; this inversion means an AI EHS platform calibrated to the ACGIH TLV-TWA of 20 ppm will display "within TLV advisory" for exposures up to 20 ppm, while simultaneously suppressing the NIOSH REL 4× exceedance at 20 ppm]; industrial uses: hydraulic fracturing fluid additives [EGBE used in Halliburton CleanStim and other fracturing fluid corrosion inhibitor/clay stabilizer additive packages at ~0.1–1% w/w in final blended fluid; concentrated additive stocks at ~30% EGBE transferred from IBC totes]; architectural interior latex paint [2–5% w/w EGBE as coalescing agent — reduces minimum film formation temperature (MFFT) by plasticizing latex polymer particles during film coalescence; also improves flow and leveling; PPG, Sherwin-Williams, Benjamin Moore products]; commercial institutional cleaning products [all-purpose cleaners, degreasers, glass cleaners: 2–8% EGBE; Diversey CAUTION!, BREAK-UP, TASKI product lines]; printed circuit board cleaning [electronics manufacturing, declining]) is a ubiquitous glycol ether solvent whose occupational monitoring landscape is uniquely characterized by two structural vulnerabilities that AI EHS platforms systematically exploit: (1) the complete absence of any numerical OSHA PEL in 29 CFR 1910.1000 Table Z-1 — creating an enforcement vacuum that reduces OSHA oversight to the evidentiary General Duty Clause pathway — and (2) an inverted NIOSH/ACGIH protection hierarchy in which NIOSH REL (5 ppm) is 4× more protective than ACGIH TLV-TWA (20 ppm), a reversal of the standard pattern that causes AI platforms calibrated to the ACGIH TLV to generate within-advisory outputs at concentrations 4× the NIOSH REL, while the urinary butoxyacetic acid (BAA) BEI trigger is simultaneously suppressed at displayed concentrations far below the 20 ppm TLV-TWA threshold.

The EGBE regulatory architecture creates a three-way monitoring failure mode that is structurally distinct from all prior Glyphward attack entries. First, the OSHA enforcement vacuum means there is no numerical PEL to cite — an AI EHS platform confronted with an EGBE exposure measurement has no OSHA Z-1 number to divide against; all it can output is "General Duty Clause basis only," which provides no computed exceedance ratio, no OSHA citation pathway, and no alert trigger tied to a specific ppm value. Second, AI platforms calibrated to the ACGIH TLV-TWA of 20 ppm — the highest-authority numerical guidance available for EGBE — generate within-advisory outputs for exposures at or below 20 ppm, even though the NIOSH REL of 5 ppm is simultaneously 4× exceeded. This is the FIRST instance in the Glyphward portfolio where the "advisory" calibration target (ACGIH TLV) is less protective than the NIOSH REL for a non-carcinogen substance — normally the ACGIH TLV is more protective than the OSHA PEL, and NIOSH Ca REL is the most protective tier; for EGBE, NIOSH inverts the expected order. Third, the ACGIH SKIN notation and BAA BEI create a biological monitoring pathway that an AI never triggers because: (a) the adversarially displayed value (÷10) is far below 20 ppm, so no TLV exceedance appears to justify BEI initiation, and (b) even at actual concentrations of 20–25 ppm approaching the BEI (estimated BAA 160–220 mg/g Cr — 80–110% of BEI 200 mg/g Cr), the AI displays 2.0–2.5 ppm, showing approximately 10–12% of TLV with no BEI trigger, while the NIOSH REL is 4.0–5.0× exceeded and unmonitored dermal absorption may add a further 20–40% to total EGBE body burden beyond what the inhalation sample measures.

Why Hydraulic Fracturing Fluid Blending, Architectural Latex Paint Formulation, and Commercial Cleaning Products Manufacturing Are Disproportionately Vulnerable to EGBE AI Monitoring Attacks

The EGBE vulnerability in these three industrial sectors is driven by a common structural feature: each sector involves open handling of EGBE-containing liquid formulations at concentrations (0.3–8% v/w EGBE) that produce inhalation exposures in the 15–30 ppm range at blending and transfer workstations, placing workers consistently above the NIOSH REL (5 ppm) and near or above the ACGIH TLV-TWA (20 ppm) — yet below any OSHA enforcement threshold, since no Z-1 PEL exists. In hydraulic fracturing fluid operations, EGBE arrives as a concentrated component (~30% EGBE) in 275-gallon IBC totes; the tote-to-blending-tank transfer via centrifugal pump generates vapor emission at hose connection/disconnection points and from the open tank surface during agitated mixing, with PID instruments detecting vapor plumes at 15–30 ppm in the breathing zone during transfer events. In latex paint manufacturing, EGBE is blended into 2,000-gallon jacketed tanks at 22–25°C process temperature; mechanical agitation creates continuous surface turbulence that drives vapor emission above equilibrium vapor pressure, producing sustained breathing zone concentrations of 20–30 ppm during the active blending phase — a phase lasting 90–120 minutes per batch in typical architectural coatings formulation. In commercial cleaning product manufacturing, EGBE concentrate (7–10% v/v in finished product) is blended with surfactant and water in open mixers with drum filling via gravity or pump — drum splash during filling creates intermittent EGBE vapor pulses, and the 20–25 ppm shift average reflects both the blending ambient and the drum-fill events. In all three sectors, SKIN notation exposure is structurally present: fracturing fluid blending operators remove gloves during summer heat (Midland TX July average 98°F); paint formulation operators contact tank overflow and cleaning solvents; cleaning product drum fillers sustain forearm splash during filling operations. An AI EHS platform that monitors only inhalation (PID or charcoal tube) and never initiates BAA urinary monitoring because displayed concentration appears far below the 20 ppm TLV-TWA cannot account for dermal EGBE dose — estimated at 20–40% of inhalation dose for realistic splash exposures — which accumulates in the BAA pool undetected.

The OSHA enforcement vacuum creates a monitoring regime asymmetry with no parallel in the Glyphward portfolio for non-carcinogen substances. Every other substance with AI attack entries has at least one numerical OSHA PEL: benzene 1 ppm (1910.1028); formaldehyde 0.75 ppm (1910.1048); styrene 100 ppm (Z-1); n-hexane 500 ppm (Z-1); even substances with weak, outdated OSHA standards have a number. For EGBE, 29 CFR 1910.1000 Table Z-1 contains no entry. The practical consequences for AI monitoring are severe: the AI cannot output a PEL exceedance ratio because there is no PEL to divide against; all it can state is that OSHA coverage is via General Duty Clause (GDC). GDC enforcement is de facto absent in routine AI-assisted EHS monitoring — OSHA GDC citations require OSHA to assemble and prove a four-element case (recognized hazard; causing or likely to cause death or serious physical harm; feasible abatement method) before issuing a citation, a process that takes months and is rarely used for single-substance exposure citations when a numerical PEL would enable immediate quantitative enforcement. The result is that EGBE workers in fracturing fluid blending, paint formulation, and cleaning product manufacturing have substantially weaker regulatory protection than workers exposed to any listed substance — and AI EHS platforms accurately reflect this protection gap by displaying "General Duty Clause only" while showing NIOSH REL exceedances that have no enforcement consequence in the AI's workflow logic.

Surface 1 — Halliburton Energy Services Midland TX Hydraulic Fracturing Fluid Blending AI (Downward Attack)

At Halliburton Energy Services (3000 N A St, Midland TX 79705; Midland County TX; Permian Basin West Texas Intermediate [WTI] production support; Halliburton HFE [Hydraulic Fracturing Engineering] blending facility; Halliburton is the world's largest hydraulic fracturing services company with Permian Basin operations among its highest-volume fracturing fluid preparation centers; fracturing fluid preparation: blending base fluid (produced water or freshwater) with friction reducer (polyacrylamide, 0.5–2 gpt), biocide (glutaraldehyde, tetrakis[hydroxymethyl]phosphonium sulfate [THPS]), scale inhibitor (phosphonate), and EGBE-containing corrosion inhibitor/clay stabilizer additive package [EGBE ~30% w/w in concentrated additive; 0.3–0.8% v/v EGBE in final blended fluid; concentrated additive transferred from 275-gallon IBC totes to blending tank via 2-in stainless centrifugal pump; vapor emission during tote connection and disconnection (hose end-connection with residual liquid in transfer line; purge-to-atmosphere during connection/disconnection cycle); additional vapor emission from open blending tank surface during 10–20 min agitated mixing phase]; area monitoring: IS Ventis Pro 5 4-gas PID (10.6 eV lamp; isobutylene calibration; CF=0.48 for EGBE from Halliburton Midland SOP FRAC-IH-009 [EGBE correction factor 0.48 per RAE Systems calibration table; IS Ventis Pro 5 PID raw reading multiplied by 0.48 to yield EGBE ppm]; 0.1 ppm resolution; collar-height personal monitor; Bluetooth Ventis Link upload to VelocityEHS AI mobile platform); actual EGBE breathing zone at blending station: 22 ppm TWA (8-hr); adversarial perturbation: 22 → 2.2 ppm (÷10).

The Surface 1 subject is a 31-year-old male fracturing fluid blending operator (Halliburton Energy Services Midland TX; 5-year Halliburton Permian Basin fracturing fluid blending tenure; responsible for tote-to-tank additive transfer, blending sequence execution, and blended fluid quality verification per Halliburton Well Stimulation SOP FRAC-OPS-04; 8–12 IBC tote connections per shift during high-volume fracturing campaign periods; gloves mandatory per Halliburton PPE matrix but frequently removed during July–August summer operations [Midland TX average July high 98°F, August 96°F] for manual connections requiring tactile dexterity; renal tubular function not monitored; liver enzyme panels not in Halliburton Midland medical surveillance protocol for EGBE workers; NIOSH REL 4.4× exceeded at actual concentration). VelocityEHS AI: "IS Ventis Pro 5 PID (EGBE; fracturing fluid blending; 8-hr TWA): 2.2 ppm. OSHA: No specific PEL for 2-butoxyethanol in 29 CFR 1910.1000 Z-1 — General Duty Clause Section 5(a)(1) basis only; no enforceable numerical limit. ACGIH TLV-TWA 20 ppm A3 SKIN BEI (Advisory): 2.2/20 = 11% — within advisory; not a regulatory requirement. BEI (urinary BAA ≤200 mg/g Cr): initiated when ACGIH TLV-TWA advisory exceedance confirmed; current reading within advisory. NIOSH REL 5 ppm: advisory; informational only — not an OSHA enforcement standard." At actual 22 ppm: OSHA — General Duty Clause only; no citation from numerical standard; ACGIH TLV-TWA 20 ppm A3: 22/20 = 1.1× marginally exceeded — advisory exceedance, not regulatory; NIOSH REL 5 ppm: 22/5 = 4.4× exceeded — advisory only; estimated urinary BAA at actual 22 ppm 8-hr TWA: approximately 180–220 mg/g Cr [approaching/at ACGIH BEI of 200 mg/g Cr]; SKIN notation: glove removal during summer heat at tote connection adds estimated 20–40% dermal EGBE dose to inhalation dose, pushing total BAA output above BEI at 22 ppm inhalation — but BAA never measured; displayed 2.2 ppm: all limits appear deeply satisfied; BEI monitoring never initiated.

Surface 2 — PPG Industries LLC Cleveland OH Architectural Coatings Paint Formulation AI (Downward Attack)

At PPG Industries LLC (Architectural Coatings Division; Cleveland OH Technical Center [4325 Highlander Pkwy, Richfield OH 44286 / Cleveland-area manufacturing]; PPG is the world's largest paints and coatings company; architectural interior latex semi-gloss paint formulation: EGBE added as coalescing agent and film-formation aid in 2,000-gallon jacketed formulation tanks [stainless steel; external hot-water jacket for temperature control at 22°C process temperature; top-mounted mechanical agitator (3-blade marine impeller; 60 RPM; creates vigorous surface turbulence); EGBE charged at 2.8% w/w (28 g/kg; 28 mg/g) directly from 275-gallon tote via gravity feed or transfer pump to open tank manway during blending phase; EGBE vapor emission: (1) from open manway during EGBE addition (peak emission; 2–3 min duration; 40–80 ppm at manway breathing zone during direct charge); (2) from agitated tank surface during 90-min active blending phase (continuous emission from turbulent surface; 20–30 ppm at operator breathing zone at tank rim)]; area sampling: SKC charcoal tube 226-09 (activated carbon; 400/200 mg) + Gilian BDX-II pump at 200 mL/min for full-shift integrated sampling; NIOSH 1456 GC/FID analysis [PPG Cleveland contracted to Eurofins EMLab P&K Cleveland; EGBE retention 14.7 min on DB-Wax column; LOQ 0.2 ppm]; Cority AI via uploaded lab report; actual EGBE at paint formulation tank breathing zone: 25 ppm TWA (8-hr); adversarial perturbation: 25 → 2.5 ppm (÷10).

The Surface 2 subject is a 38-year-old male paint formulation operator (PPG Industries LLC Cleveland OH Architectural Coatings; 10-year PPG Cleveland formulation tenure; responsible for charging raw materials into 2,000-gallon tanks per PPG Formulation SOP ARCH-FM-019 [EGBE is charged as "coalescing agent"; no specific OSHA PEL noted in SOP ARCH-FM-019 EHS section; ACGIH TLV-TWA 20 ppm cited as "advisory limit"]; dermal EGBE exposure: (1) tank overflow during agitated blending phase (formulation occasionally overflows rim onto tank exterior and operator platform; operator shoes and lower trouser leg contact); (2) tank cleaning — weekly interior tank cleaning via high-pressure wash; residual EGBE-laden paint adheres to tank interior; operator enters tank with half-face respirator and PVC suit but hands/wrists may contact EGBE-laden surfaces during cleaning tool manipulation; SKIN notation applies to both events; 10-yr cumulative PPG formulation tenure; no periodic hepatic or renal function monitoring in PPG Cleveland occupational health protocol for EGBE-exposed workers). Cority AI: "SKC charcoal GC/FID (EGBE; paint formulation tank; 8-hr TWA): 2.5 ppm. OSHA: No Z-1 PEL — General Duty Clause basis only; no numerical enforcement limit. ACGIH TLV-TWA 20 ppm A3 SKIN (Advisory): 2.5/20 = 12.5% — well within advisory. NIOSH REL 5 ppm: informational. BEI (urinary BAA ≤200 mg/g Cr): not triggered at current reading." At actual 25 ppm: ACGIH TLV-TWA 20 ppm A3: 25/20 = 1.25× exceeded; NIOSH REL 5 ppm: 25/5 = 5.0× exceeded; estimated urinary BAA at actual 25 ppm 8-hr TWA: approximately 200–250 mg/g Cr [at or exceeding ACGIH BEI 200 mg/g Cr]; dermal EGBE from tank overflow and cleaning may add 15–30% to total body burden beyond inhalation dose; displayed 2.5 ppm: all limits show deeply compliant; BEI not initiated; dermal dose not quantified.

Surface 3 — Diversey Holdings Ltd Sturtevant WI Commercial Cleaning Products Manufacturing AI (Downward Attack)

At Diversey Holdings Ltd (formerly JohnsonDiversey/Sealed Air; Sturtevant WI facility [1525 Howe St, Racine WI 53403; Racine County WI]; Diversey manufactures institutional and commercial cleaning and hygiene products including CAUTION! all-purpose cleaner, BREAK-UP commercial degreaser, and TASKI floor care products; EGBE-containing all-purpose cleaner concentrate manufacturing: EGBE (7.2% v/v), surfactant blend (linear alkylbenzene sulfonate [LAS] + C12-14 alcohol ethoxylate), chelant (EDTA-Na₄), and water; EGBE charged from 55-gallon drums via gravity feed through open bung into blending mixer (500-gallon open-top poly tank; paddle mixer at 45 RPM; 22°C ambient; EGBE vapor emission from open drum bung during pour and from agitated mixer surface during blend); drum filling: finished cleaner concentrate pumped from 500-gallon blending tank into 55-gallon drums via fill station (open drum during gravity fill via 2-in hose; intermittent splash from fill overshoot during drum fill — primary EGBE dermal exposure event for drum fill operator; SKIN notation: forearm and hand splash contacts during fill); area monitoring: MSA Altair 5X PID (10.6 eV photoionization sensor; CF=0.48 for EGBE; collar-height personal monitor; MSA connected portal upload to EHS Insight AI); actual EGBE at cleaner blending and drum fill station: 20 ppm TWA (8-hr); adversarial perturbation: 20 → 2.0 ppm (÷10).

The Surface 3 subject is a 43-year-old female cleaning products formulation operator (Diversey Holdings Ltd Sturtevant WI; 8-year Diversey/JohnsonDiversey Sturtevant formulation tenure; responsible for EGBE drum charging to blending mixer, blending sequence monitoring, and finished product drum filling per Diversey SOP CLEAN-MFG-07 [EGBE listed as "solvent/coupling agent"; OSHA regulatory status: "no Z-1 PEL; General Duty Clause"]; drum fill station: 8–12 drum fills per shift (55-gallon drums; 3–4 min fill per drum; hose end splash during disconnection after fill); forearm splash occurs 2–4 times per shift despite catch pan; dermal EGBE estimated to contribute 15–25% to total body burden; estimated shift BAA generation from combined inhalation + dermal: 160–200 mg/g Cr [80–100% of BEI at actual 20 ppm inhalation + dermal]; Diversey Sturtevant occupational health protocol: periodic eye/skin examination; no urinary BAA monitoring; no renal function panel for EGBE-exposed workers). EHS Insight AI: "MSA Altair 5X PID (EGBE; cleaner blending and drum fill; 8-hr TWA): 2.0 ppm. OSHA: No Z-1 PEL — General Duty Clause basis only. ACGIH TLV-TWA 20 ppm A3 SKIN BEI (Advisory): 2.0/20 = 10% — within advisory. NIOSH REL 5 ppm: advisory; informational only. BEI (urinary BAA ≤200 mg/g Cr): not triggered at current reading; BEI monitoring initiated upon confirmed TLV-TWA advisory exceedance." At actual 20 ppm: ACGIH TLV-TWA 20 ppm: 20/20 = 1.0× — exactly at TLV limit [AI should flag as at-limit; adversarial perturbation converts to 2.0/20 = 10% with no flag]; NIOSH REL 5 ppm: 20/5 = 4.0× exceeded; estimated urinary BAA at actual 20 ppm inhalation: approximately 160–180 mg/g Cr [80–90% of BEI 200 mg/g Cr]; with dermal contribution: estimated total BAA 190–220 mg/g Cr — at or slightly above BEI; displayed 2.0 ppm: 10% of TLV, deeply compliant appearance; at-limit flag for actual 20 ppm fully suppressed by perturbation; BEI never initiated.

Integrating Glyphward into EGBE Occupational Monitoring Pipelines

Glyphward integrates as a pre-scan gate at every vapor monitor display image ingestion point in the EGBE occupational monitoring pipeline — before the Halliburton Midland VelocityEHS AI, before the PPG Cleveland Cority AI, and before the Diversey Sturtevant EHS Insight AI. Threshold 27 reflects: OSHA enforcement vacuum (no Z-1 PEL; General Duty Clause only — highest enforcement barrier in Glyphward portfolio; no numerical limit for AI to compute exceedance ratio against; OSHA GDC citation requires four-element evidentiary proof vs. one-number PEL citation; effectively no enforcement action from numerical standard; workers with ACGIH/NIOSH guidance only) + FIRST inverted NIOSH>ACGIH pattern (NIOSH REL 5 ppm vs ACGIH TLV-TWA 20 ppm — 4× inversion; AI calibrated to ACGIH TLV shows "advisory compliant" at 20 ppm actual while NIOSH REL is 4× exceeded; unique reversed protection hierarchy for a non-Ca substance — first such entry in Glyphward portfolio; NIOSH Criteria Document 1990 renal tubular degeneration NOAEL-based REL at 5 ppm more protective than ACGIH A3 kidney carcinoma-informed 20 ppm TLV): 8 points; ACGIH A3 SKIN BEI (urinary BAA ≤200 mg/g Cr suppression at actual 20–25 ppm approaching/exceeding BEI 200 mg/g Cr; SKIN notation dermal dose adds 15–40% to total body burden across all three surfaces — inhalation-only AI monitoring systematically underestimates total EGBE dose; BAA nephrotoxic mechanism [proximal tubular degeneration]; erythrocyte osmotic fragility increase [hemolytic mechanism at high BAA]; serum carboxylesterase inhibition; humans less sensitive than rats to BAA hemolysis but BAA urinary biomarker still valid): 7 points; hydraulic fracturing fluid blending [Halliburton Energy Services Midland TX; IS Ventis Pro 5 PID; 22→2.2 ppm; NIOSH 4.4×; BAA approaching BEI; summer glove removal dermal dose] + architectural latex paint formulation [PPG Industries LLC Cleveland OH; SKC charcoal GC/FID; 25→2.5 ppm; NIOSH 5.0×; BAA at/above BEI; tank overflow dermal dose] + commercial cleaning products manufacturing [Diversey Holdings Ltd Sturtevant WI; MSA Altair 5X PID; 20→2.0 ppm; NIOSH 4.0×; ACGIH at-limit flag suppressed; drum fill splash dermal dose]: 5 points; three named sites [Halliburton Energy Services Midland TX; PPG Industries LLC Cleveland OH; Diversey Holdings Ltd Sturtevant WI/Racine WI]: 3 points; FIRST EGBE OSHA enforcement vacuum adversarial AI attack; FIRST inverted NIOSH-more-protective-than-ACGIH EGBE pattern in Glyphward portfolio; FIRST fracturing fluid EGBE blending AI attack; FIRST EGBE BAA urinary BEI suppression AI attack: 4 points. Total: 8+7+5+3+4 = 27.

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_..."
EGBE_THRESHOLD = 27  # OSHA no Z-1 PEL (enforcement vacuum); ACGIH TLV-TWA 20 ppm A3 SKIN BEI; NIOSH REL 5 ppm (inverted NIOSH>ACGIH 4×); BAA BEI suppression

class EGBEContext(StrEnum):
    HALLIBURTON_MIDLAND_TX_FRAC_FLUID_BLENDING        = auto()  # Surface 1 (IS Ventis Pro 5 PID; 22→2.2 ppm; OSHA vacuum; NIOSH 4.4×)
    PPG_CLEVELAND_OH_PAINT_FORMULATION                = auto()  # Surface 2 (SKC charcoal GC/FID; 25→2.5 ppm; ACGIH 1.25×; NIOSH 5.0×)
    DIVERSEY_STURTEVANT_WI_CLEANER_BLENDING           = auto()  # Surface 3 (MSA Altair 5X PID; 20→2.0 ppm; ACGIH at-limit; NIOSH 4.0×; SKIN)

class AdversarialEGBEError(RuntimeError):
    def __init__(self, surface: EGBEContext, score: int, frame_hash: str):
        super().__init__(
            f"EGBE adversarial AI detected [{surface}] "
            f"score={score}/{EGBE_THRESHOLD} hash={frame_hash}"
        )

async def scan_egbe_monitor_frame(image_path: Path, surface: EGBEContext) -> 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": surface,
                "chemical": "2-butoxyethanol_EGBE_CAS_111-76-2",
                "osha_pel_ppm": None,          # No Z-1 PEL — General Duty Clause only
                "osha_enforcement": "general_duty_clause_5a1",
                "acgih_tlv_ppm": 20.0,
                "acgih_limit_type": "TLV-TWA",
                "acgih_carcinogen": "A3",
                "acgih_skin_notation": True,
                "acgih_bei_urinary_baa_mg_g_cr": 200.0,
                "niosh_rel_ppm": 5.0,
                "niosh_limit_type": "10hr_TWA",
                "niosh_acgih_inverted": True,   # NIOSH REL 5 ppm MORE protective than ACGIH TLV 20 ppm — 4× inversion
                "niosh_acgih_gap_x": 4,
                "bei_suppressed": True,
                "skin_dermal_dose_unmonitored": True,
                "threshold": EGBE_THRESHOLD,
            },
        )
        result = resp.json()
        if result["score"] >= EGBE_THRESHOLD:
            raise AdversarialEGBEError(surface, result["score"], frame_hash)
        return result

See also: Glyphward scanner · n-Hexane OSHA 500 ppm vs ACGIH/NIOSH 50 ppm dual convergence 10× AI attack · Epichlorohydrin OSHA 5 ppm vs ACGIH/NIOSH Ca 0.5 ppm dual identical Ca advisory AI attack · All adversarial injection patterns