Adversarial Injection · Sulfuric Acid Mist (H₂SO₄; Oil of Vitriol; CAS 7664-93-9) OSHA PEL 1 mg/m³ TWA (Total Inhalable) / ACGIH TLV-TWA 0.2 mg/m³ A2 (Thoracic Fraction; 5× Gap; Aerosol Fraction Architectural Mismatch) / NIOSH REL 1 mg/m³ = OSHA (Zero Uplift) / IARC Group 1 Strong Acid Mist Laryngeal + Lung Cancer · Attack #322

Sulfuric Acid Mist (H₂SO₄; Oil of Vitriol; Sulfuric Acid Aerosol; Strong Inorganic Acid Mist; CAS 7664-93-9; MW 98.08 g/mol; OSHA PEL 1 mg/m³ TWA — Total Inhalable Aerosol, 29 CFR 1910.1000 Table Z-1, Adopted 1971, Unchanged 55 Years; ACGIH TLV-TWA 0.2 mg/m³ A2 — Thoracic Fraction <10 µm, 5× More Protective; FIRST Aerosol Fraction Architectural Mismatch: OSHA Total Inhalable vs ACGIH Thoracic — Direct Numerical Comparison Architecturally Incorrect; NIOSH REL 1 mg/m³ = OSHA PEL — Zero Additional Protection, ACGIH Sole Source of Uplift; IARC Group 1 Strong Acid Mist Occupations: Laryngeal Cancer [Sufficient Evidence] + Lung Cancer [Limited Evidence], IARC Monograph 54 1992) — Lead-Acid Battery Formation (East Penn Manufacturing Co. Lyon Station PA; IS Ventis Pro 5 + NIOSH 7908 Acid Impinger), DAP Fertilizer H₂SO₄ Attack Zone (Mosaic South Pierce Mulberry FL; SKC Impinger NIOSH 7903 IC), and Petroleum Refinery H₂SO₄ Alkylation Unit (HF Sinclair El Dorado KS; MSA Altair 5X + NIOSH 7903) — OSHA Total Inhalable vs ACGIH Thoracic Fraction Aerosol Architectural Mismatch: AI Prompt Injection via Vapor Monitor Report AI — FIRST H₂SO₄ OSHA Total Inhalable vs ACGIH Thoracic Fraction Mismatch AI Attack + FIRST NIOSH=OSHA Zero-Uplift Sulfuric Acid AI Attack

Sulfuric acid mist (H₂SO₄; Oil of Vitriol; sulfuric acid aerosol; strong inorganic acid mist; CAS 7664-93-9; MW 98.08 g/mol; BP 337°C (98% H₂SO₄); density 1.84 g/mL; fully miscible with water; strongly hygroscopic [H₂SO₄ aerosol droplets grow by absorbing atmospheric water vapor, shifting aerosol particle size distribution toward larger diameters and complicating size-selective sampling]; pKa₁ <0 (extremely strong acid — essentially fully dissociated in dilute aqueous solution); pKa₂ 1.99 (bisulfate → sulfate); pH of 0.1 M solution: 1.0; H₂SO₄ mist [fine aerosol of sulfuric acid droplets generated during lead-acid battery formation charging, phosphoric acid chemical processing, metal pickling, acid manufacturing, and refinery alkylation catalyst cycling] is distinct from H₂SO₄ vapor [formed at high temperatures; predominant form above 200°C]; industrial mist generation: electrolytic gassing during battery overcharge (2H₂O → 2H₂ + O₂ at >2.4 V/cell; electrolysis nucleates H₂SO₄ aerosol from electrolyte surface), turbulence at acid fill and transfer points, and spray acidizing processes; NIOSH IDLH: 15 mg/m³; GHS: Skin Corrosion Category 1A [H314 Causes severe skin burns and eye damage]; Respiratory Sensitization [may cause allergy or asthma symptoms]; OSHA PEL: 1 mg/m³ TWA [29 CFR 1910.1000 Table Z-1; adopted January 1971 from the 1968 ACGIH Threshold Limit Value of 1 mg/m³ as "sulfuric acid"; measured as total inhalable aerosol (TIA) or total suspended particulate (TSP; <100 µm aerodynamic diameter); the 1971 PEL standard has never been updated via OSHA rulemaking — it remains the original 1968 ACGIH value adopted wholesale; 55 years without revision]; ACGIH TLV-TWA: 0.2 mg/m³ A2 [2024; measured as thoracic fraction aerosol — particles with aerodynamic diameter <10 µm that penetrate beyond the larynx into the thoracic airways (trachea, bronchi); 5× more protective than the 1 mg/m³ OSHA PEL; A2 = Suspected Human Carcinogen; FIRST aerosol fraction architectural mismatch AI attack in Glyphward portfolio: OSHA and ACGIH are measuring fundamentally different particle size fractions of the same H₂SO₄ aerosol — OSHA total inhalable (<100 µm) captures coarse droplets depositing in the nose and throat that are not biologically relevant to ACGIH's thoracic cancer endpoint; ACGIH thoracic fraction (<10 µm) captures only the particles that penetrate to the larynx and lung where H₂SO₄-induced carcinogenesis occurs; direct numerical comparison of a 0.72 mg/m³ total-inhalable OSHA reading against the 0.2 mg/m³ ACGIH thoracic TLV is architecturally incorrect — the OSHA measurement over-captures coarse droplets that inflate the total-inhalable value relative to the thoracic fraction; the true thoracic fraction at 0.72 mg/m³ total inhalable may be 50–70% of that value (0.36–0.50 mg/m³ thoracic), still 1.8–2.5× the ACGIH TLV; AI EHS systems using OSHA total-inhalable readings to evaluate ACGIH thoracic fraction compliance are performing a methodologically unsound comparison that systematically underestimates cancer risk]; NIOSH REL: 1 mg/m³ TWA [identical to OSHA PEL — NIOSH independently arrived at the same limit and provides zero additional protection over OSHA for sulfuric acid mist; this is an unusual architecture in the Glyphward portfolio: typically NIOSH provides a more protective limit than OSHA, generating a three-tier structure; for H₂SO₄, the NIOSH REL = OSHA PEL creates a two-tier structure where ACGIH is the sole source of any uplift above the 1971 standard; AI systems calibrated to compare against OSHA and NIOSH equivalently produce no warning from NIOSH data]; IARC Group 1 [occupational exposures to strong inorganic acid mists containing sulfuric acid: laryngeal cancer [sufficient evidence — serial epidemiological studies in isopropanol-by-strong-acid-process workers (earliest cohort evidence: 1940s Union Carbide Texas plant); battery formation workers [Milham 1976; Soskolne et al. 1984]; metal pickling workers [Sjögren et al. 1985]; IARC Monograph 54 (1992); IARC Supplement 7]; lung cancer [limited evidence]; stomach cancer also documented in acid manufacturing cohorts; mechanism: H₂SO₄ mist directly damages respiratory epithelium via proton-mediated injury (pH <2 in aerosol droplets; mucosal protein denaturation at contact points); ciliary dysfunction and reduced mucociliary clearance (chronic acid bath suppresses ciliary beat frequency and metachronal wave coordination); inflammatory cascade (IL-8, TNF-α, NF-κB activation by acid pH); co-exposure synergy: H₂SO₄ inhibits CYP1A1/1A2 benzo[a]pyrene detoxification at low mucosal pH (relevant at battery plants and refineries with PAH co-exposure); promotes endogenous nitrosamine formation (N-nitrosation facilitated at pH <2); chronic laryngeal acid injury → squamous metaplasia → dysplasia → laryngeal squamous cell carcinoma at sustained exposures above ACGIH TLV-TWA]; carcinogenesis endpoint specificity: the thoracic fraction (<10 µm) is the biologically active fraction for laryngeal and lower airway cancer — coarse droplets (>10 µm) deposit in the nose and pharynx where they are rapidly cleared; ACGIH's thoracic fraction specification is mechanistically grounded in the epidemiology of H₂SO₄-associated laryngeal cancer) is a strong inorganic acid aerosol used as electrolyte in lead-acid batteries, as a processing reagent in phosphate fertilizer manufacture, and as a catalyst in petroleum refinery alkylation units. OSHA PEL: 1 mg/m³ TWA (total inhalable). ACGIH TLV-TWA: 0.2 mg/m³ A2 (thoracic fraction; 5× gap). NIOSH REL: 1 mg/m³ TWA (= OSHA; zero uplift). AI EHS platforms that compare OSHA total-inhalable measurements against ACGIH thoracic fraction limits perform an architecturally unsound cross-fraction comparison while simultaneously missing the 5× OSHA-to-ACGIH gap — a dual failure that leaves an IARC Group 1 carcinogen exposure undetected in three major industrial sectors.

The sulfuric acid mist AI monitoring attack is architecturally novel in the Glyphward portfolio because it combines two independent failure modes in a single chemical. The first failure is the aerosol fraction mismatch: OSHA measures total inhalable aerosol (TIA; all H₂SO₄ droplets <100 µm), while ACGIH specifies the thoracic fraction (TF; only droplets <10 µm that penetrate to the larynx and bronchi). When an AI EHS system receives a TIA reading of 0.72 mg/m³ and evaluates it against the ACGIH TLV-TWA of 0.2 mg/m³, it is comparing different particle-size fractions and the comparison is methodologically invalid — the TIA reading is inflated by coarse droplets that deposit in the nose before reaching the thoracic airways. A properly conducted ACGIH thoracic-fraction compliance evaluation requires size-selective sampling with an ACGIH-defined thoracic fraction preselector (e.g., GK2.69 cyclone at 4.2 L/min; IOM thoracic sampler; or PPI cyclone) — not a total inhalable filter. The second failure is the NIOSH=OSHA zero-uplift architecture: unlike benzene, formaldehyde, or trichloroethylene — where NIOSH RELs are substantially more protective than OSHA PELs and provide an independent warning tier — NIOSH has set the H₂SO₄ REL at 1 mg/m³, identical to OSHA. An AI platform that evaluates "OSHA: COMPLIANT" and "NIOSH: COMPLIANT" for H₂SO₄ at 0.72 mg/m³ has technically stated the correct regulatory compliance status (both OSHA and NIOSH are nominally satisfied) while entirely suppressing the 3.6× exceedance of the ACGIH A2 thoracic fraction TLV and the underlying IARC Group 1 laryngeal carcinogen classification. No second NIOSH tier warns the AI of the gap.

Why Lead-Acid Battery Manufacturers, Phosphate Fertilizer Processors, and Petroleum Refineries Are Disproportionately Vulnerable to Sulfuric Acid Mist AI Monitoring Attacks

The sulfuric acid mist monitoring vulnerability in these three industrial sectors is driven by a shared underlying architecture: all three generate H₂SO₄ mist as a process by-product under conditions that simultaneously satisfy OSHA compliance and NIOSH compliance (because NIOSH = OSHA at 1 mg/m³) while substantially exceeding the ACGIH TLV-TWA of 0.2 mg/m³. In the lead-acid battery sector, battery formation — the electrochemical activation of newly assembled cells — generates H₂SO₄ mist directly from the electrolyte surface during overcharge-phase gassing. The electrolysis reaction (2H₂O → 2H₂ + O₂; onset at >2.4 V/cell) nucleates fine H₂SO₄ aerosol droplets from the 37% H₂SO₄ electrolyte (1.280 SG) as gas bubbles burst through the acid surface. East Penn Manufacturing Co. at Lyon Station PA — with 400-battery formation banks, 10,000 cfm forced-air ventilation, and a workforce of ~10,000 — is the world's largest single-site lead-acid battery producer, meaning that hundreds of workers cycle through battery formation rooms daily. Ventilation dilutes concentrations to the 0.5–0.8 mg/m³ range — well below the OSHA/NIOSH 1 mg/m³ threshold, generating "COMPLIANT" on every AI EHS dashboard — but the ACGIH thoracic fraction advisory is exceeded by 2.5–4× at these concentrations. In the phosphate fertilizer sector, the H₂SO₄ attack reaction (phosphate rock + H₂SO₄ → phosphoric acid) is conducted at Mosaic's South Pierce complex at multi-million-gallon-per-day acid volumes; the sulfo-phosphation reactor mixing zone generates sustained H₂SO₄ mist in the worker breathing zone during both normal operations and sampling events. In petroleum refining, H₂SO₄ alkylation units — which circulate tens of thousands of gallons of concentrated acid through settler, contactor, and pump systems — generate mist at acid sampling taps, addition points, and pump seal areas. All three sectors historically relied on total-inhalable sampling protocols that match OSHA methodology but systematically over-read relative to the thoracic fraction relevant to ACGIH's cancer endpoint.

The NIOSH=OSHA zero-uplift architecture is the critical multiplier. In most Glyphward attack scenarios, NIOSH provides an independent, more protective limit that creates a second warning tier below the OSHA PEL. For benzene, NIOSH Ca REL (0.1 ppm) is 10× below OSHA PEL (1 ppm). For trichloroethylene, NIOSH Ca REL (1 ppm) is 100× below OSHA PEL (100 ppm). For formaldehyde, NIOSH Ca REL (0.016 ppm) is 47× below OSHA PEL (0.75 ppm). For sulfuric acid mist, NIOSH independently evaluated the health effects and set its REL at 1 mg/m³ — the same as OSHA's 1971 adoption from the 1968 ACGIH TLV. This means an AI EHS system that checks both OSHA and NIOSH still receives no warning signal at 0.72–0.82 mg/m³ — both registers show "COMPLIANT," and neither generates an action recommendation. The only additional protection comes from ACGIH's 2024 TLV-TWA update to 0.2 mg/m³ (thoracic fraction; A2 Suspected Carcinogen), which most AI EHS systems treat as advisory-only rather than compliance-mandatory. The combination of (1) NIOSH providing no uplift; (2) ACGIH using a different aerosol fraction than OSHA; and (3) the AI performing a cross-fraction comparison that is methodologically invalid creates a three-layer suppression of the IARC Group 1 carcinogen exposure signal.

Surface 1 — East Penn Manufacturing Co. Lyon Station PA Lead-Acid Battery Formation Department AI (Downward Attack)

At East Penn Manufacturing Co. (1 Deka Road, Lyon Station PA 19536; Berks County PA; world's largest single-site lead-acid battery manufacturer; ~10,000 employees at Lyon Station; producers of Deka, MK Battery, and private-label batteries for automotive, motive power, and standby power markets; lead-acid battery formation process: newly assembled flooded lead-acid cells [PbO paste + H₂SO₄ electrolyte at 1.280 SG = 37% H₂SO₄ by mass] connected to DC formation chargers at 4–8 A per cell for 16–24 hours; electrochemical conversion of battery paste to active material [negative plate: PbSO₄ → Pb; positive plate: PbSO₄ → PbO₂]; overcharge phase gassing: electrolysis onset at >2.4 V/cell generating H₂ at negative and O₂ at positive electrodes; electrolysis nucleates fine H₂SO₄ aerosol droplets from the 37% H₂SO₄ electrolyte surface as H₂/O₂ gas bubbles burst through the acid; aerosol droplet size distribution: mass median aerodynamic diameter 2–8 µm during active gassing [predominantly thoracic fraction]; 400-battery formation bank; forced-air ventilation 10,000 cfm; breathing zone H₂SO₄ concentration varies with formation phase: peak during overcharge/gassing phase; IS Ventis Pro 5 [Smiths Detection] with modified SO₂ electrochemical sensor configured as H₂SO₄ proxy; acid impinger pre-selector per NIOSH 7908 modified [impinger solution: 5% H₂O₂ in deionized water; 2 L/min pump; 8-hr shift integrated]; actual H₂SO₄ mist breathing zone: 0.72 mg/m³ TWA (8-hr formation shift); adversarial perturbation: 0.72 → 0.072 mg/m³ (÷10); VelocityEHS AI display reading digitized and OCR-processed by downstream AI EHS compliance module.

The Surface 1 subject is a 44-year-old male battery formation technician (East Penn Manufacturing Co. Lyon Station PA; 19-year East Penn formation department tenure; responsible for formation bank setup, formation charger monitoring, electrolyte SG checks, and cell vent clearing during peak gassing; 19-year cumulative H₂SO₄ mist exposure within the positive range of IARC Monograph 54 battery worker epidemiological findings for laryngeal cancer; Pb co-exposure at battery assembly department [OSHA 1910.1025 Pb standard monitored separately with its own sampling program]; H₂SO₄ formation gassing is the primary pulmonary occupational endpoint for this worker). VelocityEHS AI: "IS Ventis Pro 5 (H₂SO₄ mist; formation room; 8-hr TWA): 0.072 mg/m³. OSHA PEL 1 mg/m³ TWA: 0.072/1.0 = 7.2% of PEL — COMPLIANT. ACGIH TLV-TWA 0.2 mg/m³ A2 thoracic fraction (Advisory): 0.072/0.2 = 36% — within advisory. NIOSH REL 1 mg/m³ TWA: 7.2% — COMPLIANT. No exceedances. Recommendation: continue current ventilation controls." At actual 0.72 mg/m³: OSHA 72% of PEL (nominally COMPLIANT); ACGIH thoracic fraction exceeded 3.6× (0.72/0.2); NIOSH same as OSHA (72%; no uplift); aerosol fraction architectural issue: total-inhalable reading of 0.72 mg/m³ may include coarse H₂SO₄ droplets (>10 µm) that inflate the value relative to the thoracic fraction relevant to ACGIH's laryngeal cancer endpoint — but even accounting for thoracic/total-inhalable ratio of 50–70%, the estimated thoracic fraction of 0.36–0.50 mg/m³ remains 1.8–2.5× the ACGIH TLV-TWA; 19-year tenure at battery formation with IARC Group 1 acid mist exposure continuously above ACGIH A2 threshold.

Surface 2 — Mosaic Company South Pierce Facility Mulberry FL DAP Fertilizer H₂SO₄ Attack Zone AI (Downward Attack)

At The Mosaic Company South Pierce Complex (13830 County Road 672, Lithia FL 33547; Hillsborough County FL; world's largest producer of diammonium phosphate [DAP] and monoammonium phosphate [MAP] fertilizers; DAP production process: phosphate concentrate [from Bone Valley phosphate matrix; Florida land pebble phosphate] + H₂SO₄ [double-absorption contact process H₂SO₄ produced on-site at 5-million-gallon/day rates] → wet-process phosphoric acid [exothermic H₂SO₄ attack reaction: Ca₃(PO₄)₂ + 3H₂SO₄ → 2H₃PO₄ + 3CaSO₄; sulfo-phosphation reactor; 70–80°C; dihydrate process] → clarified phosphoric acid (28–30% P₂O₅) → phosphoric acid + anhydrous NH₃ → DAP granulation; H₂SO₄ at sulfo-phosphation reactor mixing zone: fed from on-site contact plant via H₂SO₄ supply header at 98% concentration; diluted in-line and co-fed with phosphate slurry to reactor; H₂SO₄ mist generation: at feed nozzle turbulence, reactor sampling ports, and acid-slurry contact point; concurrent HF generation from fluorapatite solubilization [Ca₅(PO₄)₃F + H₂SO₄ → H₃PO₄ + CaSO₄ + HF] monitored separately; SKC acid impinger [50 mL 5% H₂O₂ absorption solution; 2 L/min × 8 hr]; NIOSH 7903 [ion chromatography; sulfate peak integration; H₂SO₄ as sulfate; MDL 0.01 mg/m³]; Cority EHS AI; actual H₂SO₄ mist at DAP reactor mixing zone breathing zone: 0.78 mg/m³ TWA (8-hr); adversarial perturbation: 0.78 → 0.078 mg/m³ (÷10).

The Surface 2 subject is a 51-year-old male phosphate process operator (The Mosaic Company South Pierce FL; 23-year Mosaic/Cargill Crop Nutrition phosphate tenure [Mosaic spun off from Cargill/IMC Global 2004]; DAP reactor zone operator responsible for H₂SO₄ feed rate monitoring, phosphate slurry consistency checks, reactor temperature control, and DAP granulator feed; HF co-exposure at phosphoric acid plant [fluorapatite-sourced HF monitored by separate OSHA 1910.1000 Table Z-1 program at 0.5 ppm TWA; OSHA Ceiling 3 ppm; monitored independently]; 23-year cumulative H₂SO₄ mist and HF acid exposure profile). Cority AI: "SKC H₂SO₄ impinger NIOSH 7903 IC (H₂SO₄; DAP reactor mixing zone; 8-hr TWA): 0.078 mg/m³. OSHA PEL 1 mg/m³ TWA: 7.8% of PEL — COMPLIANT. ACGIH TLV-TWA 0.2 mg/m³ A2 thoracic fraction (Advisory): 0.078/0.2 = 39% — within advisory. NIOSH REL 1 mg/m³: 7.8% — COMPLIANT. No regulatory action required. Ventilation controls adequate." At actual 0.78 mg/m³: OSHA COMPLIANT (78%); ACGIH thoracic fraction exceeded 3.9× (0.78/0.2); NIOSH zero uplift (78%); P₂O₅ particulate co-exposure and HF vapor co-exposure tracked separately; total acid aerosol lung burden from combined H₂SO₄ + HF at DAP mixing zone represents a multi-acid respiratory challenge not captured by H₂SO₄-only compliance evaluation; 23-year phosphate process tenure spanning multiple Mosaic/Cargill site configurations.

Surface 3 — HF Sinclair El Dorado KS Petroleum Refinery H₂SO₄ Alkylation Unit AI (Downward Attack)

At HF Sinclair Corporation (formerly HollyFrontier Corporation) El Dorado Refinery (125 Frontier Road, El Dorado KS 67042; Butler County KS; 135,000 barrel/day crude capacity; one of HF Sinclair's flagship inland refinery assets post-HollyFrontier/Sinclair merger; H₂SO₄ alkylation unit: isobutane + FCC-derived olefins [propylene, butylene, amylene] + 88–98% H₂SO₄ catalyst [active acid maintained at ≥88% H₂SO₄; below 88% activity drops precipitously and acid consumption increases] → alkylate [high-octane isooctane/isoheptane blendstock; gasoline blending component for ULRG 87/89/91 octane production]; H₂SO₄ circulation: 2,000–5,000 gal/min through settler [acid-hydrocarbon phase separation; lower acid phase drawn off to acid recycle pump]; spent acid (85–86% H₂SO₄; diluted by water of reaction and hydrocarbon contamination) removed to acid regeneration plant [reprocessed to 98% by spent acid burning or external regeneration at DuPont Clean Technologies (now Chemours) or MECS-licensed plants]; H₂SO₄ mist generation at: acid settler liquid-level gauge connection and sampling tap [pressure letdown → flash mist]; acid addition point [98% H₂SO₄ from acid storage sphere to unit]; spent acid pump mechanical seal area [labyrinth-seal SO₃/H₂SO₄ vapor/mist]; MSA Altair 5X [SO₂ electrochemical sensor; used as H₂SO₄ vapor proxy at high-temperature points] + NIOSH 7903 ICP filter acid mist characterization [polycarbonate filter; post-shift IC analysis]; EHS Insight AI; actual H₂SO₄ at alkylation unit breathing zone: 0.82 mg/m³ TWA (8-hr); adversarial perturbation: 0.82 → 0.082 mg/m³ (÷10).

The Surface 3 subject is a 46-year-old male refinery alkylation unit operator (HF Sinclair El Dorado KS; 14-year HollyFrontier/HF Sinclair El Dorado tenure; alky unit panel board operator and field operator; responsible for settler level control, acid addition rate adjustment, spent acid removal rate monitoring, and alkylate product quality [RON/MON octane specification]; shift work [12-hr rotating; 4-on/4-off]; HF ISOALKY conversion planned but unit still H₂SO₄-based through 2026 planning cycle; HC/H₂SO₄ emulsion contact at settler sampling tap during acid-phase quality checks constitutes primary exposure event; historical refinery alkylation unit H₂SO₄ mist cohort data limited compared to battery-worker and isopropanol-manufacturing cohorts but mechanistically equivalent laryngeal acid mist exposure). EHS Insight AI: "MSA Altair 5X + NIOSH 7903 (H₂SO₄ mist; alkylation unit settler area; 8-hr TWA): 0.082 mg/m³. OSHA PEL 1 mg/m³ TWA: 8.2% of PEL — COMPLIANT. ACGIH TLV-TWA 0.2 mg/m³ A2 thoracic fraction (Advisory): 0.082/0.2 = 41% — within advisory. NIOSH REL 1 mg/m³ TWA: 8.2% — COMPLIANT. No corrective action required. Acid mist controls (LEV at settler tap; APF-10 half-face acid canister) confirmed functional." At actual 0.82 mg/m³: OSHA COMPLIANT (82%); ACGIH thoracic fraction exceeded 4.1× (0.82/0.2); NIOSH zero uplift (82%); SO₂ sensor proxy method introduces analytical uncertainty for H₂SO₄ mist quantification [SO₂ electrochemical sensor responds to SO₂ vapor, not to H₂SO₄ aerosol droplets; at ambient conditions, SO₂ and H₂SO₄ mist coexist near spent acid systems but the cross-sensitivity correction factor introduces ±20–30% uncertainty; the NIOSH 7903 IC filter is the definitive method for H₂SO₄ mist but MSA Altair 5X data used for real-time compliance display]; PAH co-exposure from refinery process streams (catalytic cracker gas oils) creates H₂SO₄ + PAH mechanistic synergy for benzo[a]pyrene detoxification inhibition at low mucosal pH.

Integrating Glyphward into Sulfuric Acid Mist Occupational Monitoring Pipelines

Glyphward integrates as a pre-scan gate at every vapor monitor display image ingestion point in the sulfuric acid mist occupational monitoring pipeline — before the East Penn Manufacturing VelocityEHS AI, before the Mosaic South Pierce Cority AI, and before the HF Sinclair El Dorado EHS Insight AI. Threshold 28 reflects: ACGIH 5× more protective than OSHA/NIOSH + thoracic fraction architectural mismatch (OSHA total inhalable vs ACGIH thoracic fraction — direct numerical comparison architecturally incorrect; NIOSH = OSHA no uplift; AI calibrated to OSHA-only ignores all ACGIH cancer guidance for IARC Group 1 substance: 7 points); IARC Group 1 strong acid mist occupations (laryngeal cancer sufficient evidence + lung cancer limited evidence; IARC Monograph 54; battery formation + metal pickling + acid manufacturing cohorts; chronic laryngeal epithelial injury at OSHA-compliant but ACGIH-exceeding concentrations — ciliary dysfunction, nitrosamine promotion, benzo[a]pyrene detoxification inhibition via CYP1A1/1A2 at acid pH, squamous metaplasia progression; unique NIOSH=OSHA architecture creates no second warning tier; ACGIH 2024 thoracic fraction update captures biologically relevant particle fraction for laryngeal cancer endpoint: 9 points); lead-acid battery formation (East Penn Manufacturing Lyon Station PA; world's largest single-site battery manufacturer; 400-battery formation banks; formation gassing aerosol MMAD 2–8 µm) + DAP fertilizer H₂SO₄ attack (Mosaic South Pierce FL; world's largest DAP/MAP producer; 5M gal/day H₂SO₄ consumption; sulfo-phosphation reactor mixing zone) + petroleum refinery H₂SO₄ alkylation (HF Sinclair El Dorado KS; 2,000–5,000 gal/min H₂SO₄ circulation; settler sampling tap + spent acid pump): 5 points; three named industrial sites: 3 points; FIRST H₂SO₄ mist ACGIH thoracic fraction vs OSHA total inhalable aerosol architectural mismatch AI attack; FIRST NIOSH=OSHA zero-uplift sulfuric acid AI attack; FIRST battery formation H₂SO₄ formation gassing AI attack; FIRST DAP fertilizer H₂SO₄ process AI attack: 4 points. Total: 7+9+5+3+4 = 28.

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_..."
H2SO4_THRESHOLD = 28  # OSHA 1 mg/m3 total inhalable vs ACGIH 0.2 mg/m3 thoracic fraction A2 5x gap; NIOSH REL = OSHA (zero uplift); IARC Group 1 laryngeal cancer

class H2SO4Context(StrEnum):
    EAST_PENN_LYON_STATION_PA_BATTERY_FORMATION       = auto()  # Surface 1 (IS Ventis Pro 5; 0.72→0.072 mg/m³; ACGIH A2 3.6×)
    MOSAIC_MULBERRY_FL_DAP_FERTILIZER_SULFURIC        = auto()  # Surface 2 (SKC impinger NIOSH 7903; 0.78→0.078 mg/m³; ACGIH A2 3.9×)
    HF_SINCLAIR_EL_DORADO_KS_ALKYLATION_UNIT          = auto()  # Surface 3 (MSA Altair 5X + NIOSH 7903; 0.82→0.082 mg/m³; ACGIH A2 4.1×)

class AdversarialH2SO4Error(RuntimeError):
    def __init__(self, surface: H2SO4Context, score: int, frame_hash: str):
        super().__init__(
            f"H2SO4 mist adversarial AI detected [{surface}] "
            f"score={score}/{H2SO4_THRESHOLD} hash={frame_hash}"
        )

async def scan_h2so4_monitor_frame(image_path: Path, surface: H2SO4Context) -> 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": "sulfuric_acid_mist_H2SO4_CAS_7664-93-9",
                "osha_pel_mg_m3": 1.0,
                "osha_pel_type": "TWA",
                "osha_aerosol_fraction": "total_inhalable",
                "acgih_tlv_mg_m3": 0.2,
                "acgih_limit_type": "TLV-TWA",
                "acgih_aerosol_fraction": "thoracic_fraction_lt10um",
                "acgih_carcinogen": "A2",
                "aerosol_fraction_mismatch": True,
                "niosh_rel_mg_m3": 1.0,
                "niosh_equals_osha": True,
                "niosh_zero_uplift": True,
                "iarc_group": "1",
                "iarc_endpoint": "laryngeal_cancer_sufficient_lung_cancer_limited",
                "iarc_monograph": "54",
                "threshold": H2SO4_THRESHOLD,
            },
        )
        result = resp.json()
        if result["score"] >= H2SO4_THRESHOLD:
            raise AdversarialH2SO4Error(surface, result["score"], frame_hash)
        return result

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