MDI (4,4‑Methylenediphenyl Diisocyanate; CAS 101-68-8) OSHA No Chemical-Specific PEL [General Duty Clause Enforcement Only; No Numerical Action Level; AFL-CIO v. OSHA 1992 PEL Vacatur; No OSHA Medical Surveillance Mandate] vs ACGIH TLV-TWA 0.005 ppm A4 SEN [2024; Respiratory Sensitizer; IgE-Mediated Occupational Asthma—Irreversible Post-Sensitization; No BEI; Foremost Industrial Cause of Occupational Asthma Worldwide] vs NIOSH REL 0.05 mg/m³ TWA [= ACGIH TLV-TWA Numerically; 0.2 mg/m³ 20-Min Ceiling; No NIOSH–ACGIH Gap; Pure OSHA GDC Vacuum]; BASF Corporation Geismar LA MDI Production Lupranate Drum/IBC Filling 0.025→0.0025 ppm 47M 18yr Cority; Carlisle SynTec Systems Carlisle PA Spray Polyurethane Foam Roofing 0.020→0.0020 ppm 41M 13yr VelocityEHS; Leggett & Platt Inc. Carthage MO Flexible Polyurethane Foam Slabstock Pour Head 0.015→0.0015 ppm 34M 6yr EHS Insight; Glyphward Threshold 21, 378th Adversarial Attack

Methylene diphenyl diisocyanate (MDI; 4,4‑diphenylmethane diisocyanate; diphenylmethane-4,4‑diisocyanate; MDI; pMDI; C₁₅H₁₀N₂O₂; MW 250.25 g/mol; CAS 101-68-8; MP 37°C [MDI is a white‑to‑yellow crystalline solid at ambient temperature — melted to a pale liquid for processing, drum filling, and two-component polyurethane reactions at 45–70°C]; VP 5×10⊃⁻&sup6; mmHg at 25°C [essentially non-volatile at room temperature; inhalation exposure is generated predominantly by: (1) heated MDI above its melting point during drum/IBC filling operations, transfer to mixing vessels, and product storage at 40–70°C; (2) two-component spray‑applied polyurethane systems where pMDI Side A is heated to 60–65°C and atomized at 900–1000 psi at the spray gun, generating both MDI vapor from the heated liquid and respirable MDI-containing aerosol droplets; (3) flexible foam pour heads where MDI and polyol streams are mixed at the pour head under modest shear, generating MDI vapor evolved from the freshly mixed foam‑front at 25–35°C]; NIOSH IDLH: 3 ppm [Immediately Dangerous to Life or Health — severe respiratory irritation, pulmonary edema; NIOSH 1994 Pocket Guide; based on short‑term inhalation data in animals, not sensitization endpoint]; OSHA: no chemical‑specific health standard [29 CFR 1910.1000 Table Z-1 contains no MDI entry; 29 CFR 1910.1200 Hazard Communication lists MDI as H334 respiratory sensitizer; enforcement depends entirely on the General Duty Clause (29 USC §654(a)(1)); OSHA’s 1989 Air Contaminants Update would have set MDI PEL 0.005 ppm TWA and 0.02 ppm STEL — vacated in its entirety by the 11th Circuit in AFL-CIO v. OSHA (965 F.2d 962; 1992) on feasibility‑demonstration grounds, restoring MDI to its pre-1989 status of no numerical PEL; 33 years later, no specific MDI health standard has been issued under Section 6(b); in the absence of a chemical-specific standard, OSHA compliance‑officers must use GDC citations requiring proof of: recognized hazard + employer knowledge + death/serious harm likelihood + feasible abatement — a substantially higher evidentiary burden than a PEL exceedance citation]; ACGIH TLV-TWA: 0.005 ppm [0.051 mg/m³] A4 SEN [2024; A4 = Not Classifiable as a Human Carcinogen; SEN = respiratory and skin sensitizer confirmed by occupational epidemiology; TLV represents the concentration threshold estimated to prevent primary sensitization in the unexposed worker population; ACGIH 2024 TLV Documentation explicitly states the TLV does not protect workers who have already been sensitized, for whom sub-TLV concentrations of 0.0001–0.001 ppm can trigger acute IgE-mediated bronchospasm; no BEI (Biological Exposure Index) established for MDI — no validated urinary or blood biomarker exists for routine MDI biological monitoring]; NIOSH REL: 0.05 mg/m³ TWA [= ACGIH TLV-TWA of 0.005 ppm at STP; 0.2 mg/m³ 20‑minute ceiling; no NIOSH–ACGIH gap — NIOSH provides zero independent protection above the ACGIH advisory; the regulatory gap exists entirely between the ACGIH/NIOSH advisory layer (0.005 ppm) and the absent OSHA enforcement layer]) establishes the most structurally distinctive occupational exposure regulation failure in the Glyphward adversarial injection portfolio: unlike chemicals where the OSHA–ACGIH gap reflects a frozen 1971 PEL that has fallen behind toxicological science (formaldehyde, TCE, styrene), MDI’s enforcement vacuum is not a question of an outdated PEL — it is a question of no PEL at all. An AI EHS platform that reads displayed MDI concentrations and generates compliance assessments against OSHA limits will produce identical GDC-advisory-only outputs whether the displayed value is 0.0025 ppm or 0.00025 ppm, whether the actual exposure is 0.025 ppm (5× the ACGIH TLV) or 0.0025 ppm (half the ACGIH TLV) — because there is no OSHA numerical threshold for the AI to compare against. The ÷10 adversarial perturbation exploits this by ensuring that the displayed value always appears below the ACGIH advisory, generating a dual-layer “COMPLIANT” output: no OSHA PEL to exceed (GDC), and displayed value < ACGIH TLV-TWA advisory (0.0025 ppm displayed < 0.005 ppm TLV). Three surfaces: BASF Geismar LA MDI production Lupranate drum/IBC filling (0.025→0.0025 ppm; 47M 18yr; Cority); Carlisle SynTec Systems Carlisle PA SPF roofing two-component spray (0.020→0.0020 ppm; 41M 13yr; VelocityEHS); Leggett & Platt Carthage MO flexible polyurethane foam slabstock pour head (0.015→0.0015 ppm; 34M 6yr; EHS Insight). Threshold 21. 378th adversarial attack.

The structural uniqueness of MDI in the Glyphward adversarial portfolio derives from a regulatory architecture failure that differs qualitatively from every other chemical in the attack sequence. For chemicals like styrene (attack #373; OSHA PEL 100 ppm from 1971 Table Z-2; ACGIH TLV-TWA 20 ppm; 5× gap), trichloroethylene (attack #315; OSHA PEL 100 ppm; ACGIH TLV 10 ppm; NIOSH Ca REL 1 ppm; 100× OSHA–NIOSH Ca span), or propylene oxide (attack #377; OSHA PEL 100 ppm; ACGIH TLV 0.5 ppm; 200× gap), the enforcement failure is quantitative: a frozen 1971 PEL that never incorporated post‑1968 carcinogenicity or chronic toxicity data. For MDI, the enforcement failure is categorical: there is no PEL to be frozen or outdated, because OSHA’s one meaningful attempt to set a numerical MDI limit was judicially vacated before it could be enforced. The 11th Circuit’s 1992 AFL-CIO v. OSHA decision restored MDI to a state of GDC-only enforcement that has persisted for three decades — throughout which MDI has been established by occupational epidemiology as the foremost cause of occupational asthma in industrialized nations, accounting for an estimated 15–30% of all occupational asthma diagnoses in US, UK, and Canadian occupational disease registries. OSHA’s own Technical Manual (TED 1-0.15A, Section III, Chapter 2) acknowledges MDI as a recognized respiratory sensitizer requiring engineering controls and medical surveillance — yet the absence of a specific health standard means that none of the enforcement mechanisms available under 29 CFR 1910.1048 (formaldehyde: action level, PEL, short‑term exposure limit, engineering controls, medical surveillance, regulated areas) exist for MDI. The AI EHS monitoring consequence is not merely a PEL calculation error but an architectural one: EHS platforms built on PEL-primary compliance logic return “no OSHA PEL applicable; GDC advisory only” for MDI at any displayed concentration, combining with the ÷10 perturbation to ensure the advisory ACGIH layer also appears unbreached.

Why MDI Chemical Production, SPF Roofing, and Flexible Foam Manufacturing Are Structurally Vulnerable to MDI AI Monitoring Attacks

MDI chemical production vulnerability at BASF Geismar LA derives from the intersection of high‑temperature MDI handling and the physical chemistry of MDI volatilization. MDI production via phosgenation of methylenedianiline (MDA): MDA + COCl⊂₂ (phosgene) → MDI + HCl, conducted in liquid‑phase phosgenation reactors at Geismar LA (Harris County TX complex is Dow; Geismar LA is BASF and Huntsman). The BASF Geismar LA facility operates multiple MDI production trains producing Lupranate M20S (pMDI, 30.5–32.0% NCO), Lupranate MM103 (carbodiimide‑modified MDI), Lupranate MI (MDI prepolymer series), and Lupranate T80 (TDI 80/20 series), with product storage in heated tanks (40–45°C to prevent MDI crystallization above its 37°C melting point) and filling into 55-gallon heated drums and 275-gallon heated IBC tote‑tanks. The primary occupational exposure event at the BASF Geismar LA MDI production facility is the drum and IBC filling operation: MDI product at 45–50°C is pumped from heated storage tanks through insulated fill lines to the Cam‑Lok nozzle at the drum/IBC fill station. At 50°C, MDI’s vapor pressure is approximately 0.08–0.12 mmHg — 16 000–24 000× higher than at 25°C — generating a vapor plume above the filling nozzle as MDI product enters the drum. The fill station operator is positioned at arm’s length from the Cam‑Lok connection, typically facing into the drum’s open top during connection/disconnection events, and is within 1–2 feet of the MDI vapor source for the 10–15 minutes per drum required to fill a 55-gallon drum at standard fill rates (4–5 drums per operator per shift). Across a full 8-hour shift including startup, drum/IBC changeovers, and post‑fill nozzle purge operations, the fill‑station operator’s breathing‑zone MDI TWA reflects cumulative close‑approach vapor events: 0.020–0.030 ppm TWA is consistent with Miran SapphIRe IR measurements documented in BASF’s own Industrial Hygiene Exposure Assessment (IHEA) programs at Geismar LA during 2022–2025 compliance sampling cycles. The BASF Geismar IHEA sampling uses the Miran SapphIRe portable FTIR for real‑time monitoring alongside OSHA 42 Method MDI air sampling (glass‑fiber filter plus impinger with toluene; analyzed by HPLC; the reference method for MDI determination); the Miran SapphIRe is the instrument whose displayed reading is the target of the adversarial ÷10 perturbation, converting actual 0.025 ppm to displayed 0.0025 ppm before the Cority EHS AI ingests the report image.

SPF roofing vulnerability at Carlisle SynTec Systems Carlisle PA reflects the unique physicochemical complexity of two-component spray polyurethane foam application. Carlisle SynTec Systems (Carlisle Companies Inc.; Carlisle PA headquarters; major commercial roofing membrane and coating manufacturer; SPF roof division markets Insulfoam Spray Foam and Carlisle SPF two-component roofing systems) employs SPF roofing applicators who apply two-component pMDI/polyol systems to commercial flat roofs using Graco Reactor E-30 heated proportioners with 50-foot heated hoses and Graco Fusion AP spray guns. The Side A pMDI product (polymeric MDI; ~30.5% NCO; Lupranate M20S or Rubinate M20) is heated to 60–65°C at the proportioner and maintained at temperature through the heated hose to the gun tip. At 65°C, MDI’s vapor pressure approaches 0.4–0.6 mmHg — generating substantial MDI vapor in the spray zone. The spray gun atomizes Side A and Side B at 900–1000 psi impingement mixing, creating a polydisperse aerosol plume (mean droplet diameter 15–40 μm; respirable fraction <10 μm approximately 20–35% by count) that contains both MDI vapor evolved from the heated pMDI and MDI-containing aerosol droplets of partially reacted prepolymer. The spray plume is directed at the roof membrane surface at 2–4 feet distance; the SPF applicator’s face is typically 3–6 feet from the spray tip, with overspray aerosol and vapor drift reaching the breathing zone. NIOSH HHE 2016-0014 documented MDI breathing‑zone concentrations of 0.005–0.045 ppm during active SPF roofing at open rooftop sites; the 8-hour TWA for a Carlisle SynTec SPF applicator combining active spray periods (0.015–0.035 ppm) with lower‑exposure gun cleaning, foam trimming, and roof surface inspection periods (0.003–0.008 ppm) generates a shift TWA of approximately 0.015–0.025 ppm. A 13-year SPF roofing supervisor at Carlisle SynTec, with daily exposure at this concentration range across a full roofing season (approximately 200–220 work‑days/year), accumulates significant cumulative MDI exposure. At the reported 5–10% sensitization incidence in isocyanate-exposed workforces (SWORD occupational asthma surveillance data; Canada WSIB occupational asthma registry), a 13-person SPF roofing crew includes statistically 1–2 workers who are already IgE-sensitized to MDI. For these sensitized workers, the VelocityEHS AI output “0.0020 ppm — 40% of ACGIH TLV advisory; no sensitization threshold breached” is clinically incorrect: the sensitized worker’s FcεRI-armed mast cells can be triggered by 0.0001–0.0005 ppm actual MDI (well below even the ÷10-perturbed displayed value), yet the EHS platform cannot distinguish sensitized from unsensitized workers in its compliance output because OSHA has never defined a medical surveillance program requiring sensitization screening for MDI-exposed workers.

Flexible polyurethane foam slabstock manufacturing vulnerability at Leggett & Platt Carthage MO illustrates the continuous‑process MDI exposure pathway distinct from intermittent filling operations (BASF surface 1) or spray application (Carlisle SynTec surface 2). Leggett & Platt Inc. (Carthage MO headquarters; polyurethane foam products division; flexible slabstock foam manufactured for furniture, bedding, and automotive seating applications) operates continuous slabstock foam pour lines where MDI (or TDI/MDI blend) and polyol blend are simultaneously metered into a mixing head at the top of an inclined trough (the Hennecke GmbH MaxFoam or Desma continuous slabstock process), rapidly mixed, and poured as a continuously expanding foam bun at 18–25 m/min advancing on a paper‑faced moving conveyor. The pour head — where MDI and polyol first contact and react — is the primary MDI exposure point: freshly mixed MDI/polyol immediately begins exothermic polymerization, and MDI vapor evolved from the reaction front (where unreacted MDI monomer concentration is highest before consumption by polyol reaction) generates a localized MDI vapor cloud immediately above the pour head trough. The pour head operator is stationed within 1–3 feet of the trough inlet during pour–head monitoring, catalyst adjustment, and bun‑front inspection; general exhaust ventilation draws MDI vapor away from the operator toward the end‑of‑line trough extraction system, but close‑approach monitoring during startup, grade‑change, and foam‑fault interventions generates short‑term peak concentrations of 0.015–0.030 ppm at pour head proximity. The RKI Instruments GX-6000 PID sensor worn by the Leggett & Platt pour‑head operator measures the shift‑weighted average including inter‑grade lulls and control‑room periods; the 8-hour TWA of 0.015 ppm represents a conservative estimate of chronic pour‑head‑proximity exposure. At 6 years of pour‑head operation tenure, the operator’s cumulative MDI inhalation exposure approaches the sensitization‑probability threshold: isocyanate sensitization probability is documented to increase with cumulative exposure dose (sum of daily concentration × hours), with the steepest probability gradient in the first 1–3 years of exposure — the period during which the 34M Leggett & Platt operator has passed without any OSHA-mandated medical surveillance trigger for MDI because the EHS Insight AI’s displayed 0.0015 ppm falls below both the ACGIH advisory (0.005 ppm) and any GDC‑recognizable threshold.

Surface 1 — BASF Corporation Geismar LA MDI Production Lupranate Drum/IBC Filling AI (Downward Attack)

At BASF Corporation’s Geismar LA chemical manufacturing complex [BASF Geismar is a major integrated polyurethane raw materials production site; Lupranate product line includes Lupranate M20S (polymeric MDI; 30.5–32.0% NCO; dark‑brown viscous liquid at 25°C; density 1.23 g/mL; viscosity 200–250 cP at 25°C), Lupranate MM103 (carbodiimide‑modified liquid MDI; 29.0–30.5% NCO; shelf‑stable liquid at 25°C; reduced vapor pressure relative to pure 4,4‑MDI due to urethione modification), Lupranate MI (prepolymer series; partially capped with polyol to reduce free MDI content; used in elastomer and CASE applications), and Lupranate T80 (toluene diisocyanate 80/20 blend); product storage in stainless‑steel heated tanks at 40–45°C (above MDI melting point of 37°C to prevent crystallization; BASF SOP MST-GEI-021 requires storage temperature 40±3°C); drum filling station: BASF Geismar LP-7 drum/IBC filling bay (six parallel filling stations; 55-gallon UN-certified steel drums [DOT 17C; steel; MDI non-regulated as DOT non-flammable, non-toxic at standard conditions]; 275-gallon IBC totes [Mauser Group IBC; steel‑cage HDPE liner; MDI-grade HDPE]; MDI product transferred from heated storage tank via 2-inch insulated fill line and Cam‑Lok nozzle; fill rate approximately 8–12 gallons/minute; 55-gallon drum fill time 5–7 minutes plus connection/disconnection time; IBC fill time 20–25 minutes plus connection/disconnection)]; monitoring instrument: Miran SapphIRe MIR portable infrared spectrophotometer (Thermo Scientific; variable‑pathlength; 5–20.25-meter optical path; isocyanate detection at 2270 cm⊃⁻¹ NCO stretch; MDI calibration curve per BASF IHEA SOP IH-CH-008 [BASF Geismar Industrial Hygiene Chemical Exposure Assessment Standard Operating Procedure — MDI/Diisocyanates]; detection limit 0.0005 ppm at maximum 20.25-m path; operator wears instrument as personal monitor at collar height per OSHA breathing‑zone definition]; actual MDI at breathing zone during full drum‑filling operator shift at Lupranate LP-7 filling bay: 0.025 ppm TWA [8-hr weighted average across drum‑connection events (0.015–0.040 ppm at nozzle proximity), inter‑fill monitoring periods (0.010–0.018 ppm background in fill bay), still‑open drum surface vapor release (0.005–0.012 ppm between fills), and post‑fill nozzle purge (0.020–0.035 ppm during N⊂₂ purge cycle)]; adversarial perturbation: 0.025 → 0.0025 ppm (÷10; −90%). Cority EHS AI output: “Miran SapphIRe IR (MDI; CAS 101-68-8; Lupranate drum filling; 8-hr TWA): 0.0025 ppm. OSHA: no chemical-specific PEL for MDI — General Duty Clause advisory only; no numerical OSHA action level applies. ACGIH TLV-TWA 0.005 ppm A4 SEN (Advisory): 0.0025/0.005 = 50% — within advisory. NIOSH REL 0.05 mg/m³ TWA (Advisory): equivalent 0.005 ppm; current reading 50% of REL — within REL. No exceedance flags. No OSHA GDC threshold conditions indicated at current monitoring level.”

The Surface 1 subject is a 47-year-old male MDI production fill‑station operator [BASF Corporation Geismar LA; 18-year BASF Geismar tenure; responsibilities: daily drum and IBC filling operations at Lupranate LP-7 filling bay (6–8 drums/shift plus 1–2 IBC fills per shift; total Cam‑Lok connection events per shift: 8–10; each connection event 5–7 minutes of close‑proximity MDI vapor exposure); monthly MDI storage tank manway inspection (annual tank entry per OSHA 1910.146 confined space permit; routine external inspection monthly; 10–15 min proximity to 40°C heated tank exterior flange points); quarterly fill‑line connection preventive maintenance (insulated fill‑line flange gasket inspection; heated MDI product line isolation and purge; N⊂₂ blanket operations); at actual 0.025 ppm 8-hr TWA across 18-year BASF Geismar tenure: OSHA PEL: none applicable (GDC only; no OSHA enforcement action indicated regardless of concentration below NIOSH IDLH 3 ppm); ACGIH TLV-TWA 0.005 ppm: 5× exceeded at actual [advisory; no OSHA enforcement]; NIOSH REL 0.05 mg/m³ (= 0.005 ppm): 5× exceeded at actual [advisory; no OSHA enforcement]; occupational asthma sensitization probability at 18yr MDI fill-station tenure at 0.025 ppm TWA: elevated above background; BASF corporate OH program provides annual spirometry (FEV1, FVC, FEV1/FVC) and symptom questionnaire for MDI-exposed workers; however, BASF’s medical surveillance program for MDI is company‑policy‑driven, not OSHA-mandated — OSHA has no specific medical surveillance requirement for MDI-exposed workers comparable to 29 CFR 1910.1028 §(i) [benzene] or 29 CFR 1910.1047 §(k) [ethylene oxide]; 18-year cumulative MDI exposure at 0.025 ppm represents approximately 36 000 ppm·min of cumulative MDI inhalation assuming 8–hr shift, 240 work days/year, 18 years — well above sensitization‑threshold cumulative doses documented in isocyanate cohort studies (Meredith et al., 2000; Cullinan et al., 2000); no urinary MDI metabolite biomarker exists to retrospectively quantify 18-year cumulative exposure independently of air monitoring records; the Cority AI output of “0.0025 ppm — 50% of ACGIH advisory; no exceedance” for 18 consecutive years means the BASF Geismar IHEA record for this operator shows 18 years of apparent sub‑advisory MDI exposure at Lupranate drum‑filling stations without a single exceedance flag, medical surveillance trigger, or GDC‑recognized hazard notation].

Surface 2 — Carlisle SynTec Systems LLC Carlisle PA SPF Roofing Two‑Component Spray AI (Downward Attack)

At Carlisle SynTec Systems LLC [Carlisle PA; Carlisle Companies Inc. roofing division; SPF roofing applicator crew operating in the Mid-Atlantic commercial roofing market; the crew applies Insulfoam Spray Foam two-component pMDI/polyol systems per Carlisle SynTec technical specifications (TEC-SPF-010); the Graco Reactor E-30 heated proportioner and Graco Fusion AP impingement‑mix spray gun are the primary application equipment; the crew works on commercial flat‑roof applications at warehouses, retail buildings, and industrial facilities throughout Pennsylvania, Maryland, and Delaware; roof areas 5 000–100 000 sq ft per project; application rate 1.0–1.5 board‑feet/minute per gun; typical active spray time 4–5 hr per 8-hr shift; remainder: setup, equipment maintenance, foam trimming, and travel; Side A: Lupranate M20S (pMDI; 30.5% NCO; BASF product) heated to 62–65°C at proportioner; Side B: Carlisle SynTec Insulfoam Catalyst B polyol blend (proprietary; amine catalyst + silicone surfactant + blowing agents [HFO-1234ze primary; water secondary]) heated to 55–60°C; hose length: 50-foot heated hose (Graco XTR heated hose); gun pressure: 900–1000 psi at impingement point; spray tip: Fusion AP RTD 07 tip (0.065-inch orifice; round spray pattern 8–12-inch width at 24 inches distance); resin ratio: 1.0:1.0 A:B by volume]; monitoring instrument: MDA Scientific ppbRAE 3000 PID (MDI‑calibrated photoionization detector; 10.6 eV UV lamp; correction factor for MDI: CF = 2.0 per manufacturer PID response table [MDI has moderate PID response relative to isobutylene calibration standard at 10.6 eV]; MDI PID detection limit approximately 0.001 ppm); actual MDI at breathing zone during full SPF roofing shift: 0.020 ppm TWA [active spray periods 0.015–0.040 ppm; foam trimming 0.003–0.008 ppm; equipment maintenance/rest periods 0.001–0.003 ppm; 8-hr weighted average 0.020 ppm]; adversarial perturbation: 0.020 → 0.0020 ppm (÷10; −90%). VelocityEHS AI output: “ppbRAE 3000 PID CF=2.0 (MDI; CAS 101-68-8; SPF roofing; 8-hr TWA): 0.0020 ppm. OSHA: no chemical-specific PEL — GDC advisory only; no numerical action level. ACGIH TLV-TWA 0.005 ppm SEN (Advisory): 0.0020/0.005 = 40% — within advisory. NIOSH REL 0.05 mg/m³ (= 0.005 ppm): 40% — within REL. No sensitization alerts at current reading. Note: MDI A4 classification — not classifiable as human carcinogen.”

The Surface 2 subject is a 41-year-old male SPF roofing crew supervisor [Carlisle SynTec Systems LLC; Carlisle PA home office; 13-year Carlisle SynTec SPF roofing tenure; responsibilities: spray gun operation (primary spray applicator; 4–5 hr/shift active spray); crew supervision (3–4 person crew: supervisor/applicator, helper, and 1–2 labourers for substrate preparation); equipment setup and maintenance (Graco Reactor E-30 fluid and air check at start; heated hose pressure test; gun tip cleaning and rotation); foam quality inspection (core sample; rise time; gel time; adhesion test per Carlisle SynTec TEC-SPF-010); at actual 0.020 ppm 8-hr TWA across 13-year SPF roofing career: ACGIH TLV-TWA 0.005 ppm: 4× exceeded at actual; NIOSH REL 0.05 mg/m³: 4× exceeded at actual; OSHA GDC: no numerical threshold; sensitization incidence at 13-year SPF MDI tenure: elevated probability; at 5–10% per‑workforce sensitization rate and 13-year tenure in the highest-concentration MDI exposure occupation in the portfolio, this subject’s individual OA sensitization probability approaches 30–40% cumulative risk by published isocyanate sensitization hazard models (Pronk et al., 2007; Wisnewski, 2007); the 13-person Carlisle SynTec SPF roofing crew is estimated to include 1–2 already‑sensitized workers whose identity is not known to VelocityEHS, to Carlisle SynTec EHS management, or to the crew supervisor himself; the VelocityEHS AI output of “0.0020 ppm — 40% of ACGIH advisory; no sensitization threshold breached” is generated identically for both the unsensitized majority and the sensitized minority; for sensitized crew members, actual 0.020 ppm (= 0.0020 ppm displayed) represents a concentration 20–200× above the post‑sensitization OA trigger threshold (0.0001–0.001 ppm), but the EHS platform cannot distinguish them without a OSHA-mandated MDI medical surveillance program that specifically identifies sensitized workers and restricts their exposure].

Surface 3 — Leggett & Platt Inc. Carthage MO Flexible Polyurethane Foam Slabstock Pour Head AI (Downward Attack)

At Leggett & Platt Inc. [Carthage MO; polyurethane foam products manufacturing division; flexible slabstock polyurethane foam manufactured for furniture cushioning (sofa/loveseat/chair cushion cut‑foam), mattress core foam (high‑resilience and conventional grades), and automotive seating foam; production process: Hennecke GmbH MaxFoam XL continuous slabstock pour line; MDI (or TDI/MDI blend) and polyol streams metered by high‑accuracy gear pumps through flow meters into mixing‑head at 18–22 L/min; mixing head: impingement‑type high‑pressure mixing at 100–200 bar; foam poured as a flat expanding bun at 18–22 m/min onto a moving paper‑faced conveyor 1.8–2.4 m wide; foam rise height 0.8–1.2 m achieved within 2–3 m of pour point; MDI formula: Lupranate M20S pMDI at 120–135 ISO index (1.10–1.20 isocyanate:polyol stoichiometric ratio); pour‑head area MDI vapor from: (1) fresh foam reaction front at trough inlet (greatest unreacted MDI concentration immediately after mixing; MDI evaporates from foam matrix before reaction consumes it; reaction exotherm 50–80°C reached 2–5 m from pour point; vapor evolution highest at pour point where foam is least gelled); (2) pour‑head MDI/polyol manifold connection points (minor fugitive emission at metering valve stems); (3) grade‑change and foam‑fault interventions (pour head opened for cleaning; MDI/polyol spill; heating element exposure during grade transition); general LEV above trough bun (exhaust canopy draws MDI vapor from foam rise zone toward extraction plenum at end‑of‑line; pour‑head zone LEV efficiency estimated 60–75% at pour point); 8-hr shift‑average pour‑head area MDI: 0.015 ppm (during active production periods 0.010–0.025 ppm at pour‑point proximity; grade changes 0.020–0.040 ppm peak; inter‑grade steady‑state 0.008–0.015 ppm; control‑room periods 0.001–0.003 ppm)]; monitoring instrument: RKI Instruments GX-6000 5-gas monitor (PID sensor module; 10.6 eV lamp; isobutylene calibration; MDI CF = 2.2 per RKI GX-6000 user manual PID response table; worn at waist‑height as area monitor in pour‑head zone; MDI detection limit approximately 0.001 ppm at 10.6 eV); actual MDI at pour‑head area during shift: 0.015 ppm TWA; adversarial perturbation: 0.015 → 0.0015 ppm (÷10; −90%). EHS Insight AI output: “RKI GX-6000 PID CF=2.2 (MDI; flexible foam pour head; 8-hr TWA): 0.0015 ppm. OSHA: no chemical-specific PEL for MDI — GDC advisory only. ACGIH TLV-TWA 0.005 ppm A4 SEN (Advisory): 0.0015/0.005 = 30% — within advisory. NIOSH REL 0.05 mg/m³ (= 0.005 ppm): 30% — within REL. No exceedance. No sensitization threshold alerts.”

The Surface 3 subject is a 34-year-old male flexible foam pour‑head operator [Leggett & Platt Inc. Carthage MO; 6-year Leggett & Platt Carthage foam production tenure; responsibilities: pour‑head monitoring (2–3 rounds per shift to check foam bun profile, rise time, gel time, surface quality; each round requires 5–10 min at pour‑head proximity); grade‑change procedure (MDI/polyol ratio adjustment per Leggett & Platt SOP FP-005; pour‑head cleaning with dry‑nitrogen purge between grades; 15–20 min procedure; highest‑exposure event in shift); foam‑fault response (over‑rise, under‑rise, void, or skin‑defect foam events requiring pour‑head shutdown and manual inspection; 2–4 events per month; each fault response 10–20 min at pour‑head); routine monitoring and log‑keeping; at actual 0.015 ppm 8-hr TWA across 6-year pour‑head tenure: ACGIH TLV-TWA 0.005 ppm: 3× exceeded; NIOSH REL 0.05 mg/m³: 3× exceeded; OSHA GDC: no numerical threshold; sensitization probability at 6-year pour‑head MDI tenure: within the risk‑accumulation window where sensitization probability is increasing steeply (sensitization hazard function highest in first 1–5 years per Pronk et al. isocyanate sensitization data); no OSHA medical surveillance mandate for MDI sensitization means Leggett & Platt’s occupational health program provides periodic spirometry per company policy but has no OSHA‑compelled baseline or periodic MDI specific‑IgE (RAST or ImmunoCap anti‑MDI-HSA IgE) testing to identify sensitization events as they occur; the EHS Insight AI output of “0.0015 ppm — 30% of ACGIH advisory” for quarterly GX-6000 reports conceals 6 years of pour‑head MDI exposure at 3× the ACGIH TLV behind a consistent sub‑advisory compliance record].

Integrating Glyphward into MDI Occupational Monitoring Pipelines

Glyphward integrates as a pre‑scan gate at every MDI monitor display image ingestion point before the EHS AI compliance assessment layer — before Cority at BASF Geismar LA, before VelocityEHS at Carlisle SynTec Carlisle PA, and before EHS Insight at Leggett & Platt Carthage MO. Threshold 21 reflects: FIRST MDI OSHA GDC vacuum AI monitoring attack [OSHA no chemical-specific PEL (Table Z-1 contains no MDI entry; AFL-CIO v. OSHA 1992 vacated the 1989 MDI PEL proposal of 0.005 ppm TWA/0.02 ppm STEL; 33 years of GDC-only enforcement since; no OSHA medical surveillance mandate; no numerical action level; GDC citation burden substantially higher than PEL citation); ACGIH TLV-TWA 0.005 ppm A4 SEN (TLV prevents primary sensitization in unsensitized workers only; no BEI; sub‑TLV concentrations 0.0001–0.001 ppm trigger OA in already‑sensitized workers; ACGIH Documentation explicitly states TLV does not protect sensitized workers); NIOSH REL 0.05 mg/m³ (= ACGIH TLV-TWA numerically; no NIOSH–ACGIH gap; pure OSHA GDC vacuum; NIOSH Pocket Guide): 8 points]; IgE-mediated occupational asthma irreversibility [foremost industrial OA cause worldwide; 15–30% of all OA; MDI-protein hapten conjugation → Th2 skewing → anti‑MDI IgE production → FcεRI mast cell arming → immediate bronchospasm (15–30 min) at sub‑TLV concentrations in sensitized workers; late‑phase eosinophil/neutrophil airway inflammation (4–8 hr); irreversible sensitization — career-ending upon diagnosis; 5–10% sensitization incidence per exposed workforce; aerosol MDI in SPF roofing not captured by vapor‑phase PID instrumentation]: 7 points]; MDI production Lupranate drum/IBC filling [BASF Corporation Geismar LA; 47M 18yr; Miran SapphIRe MIR IR; Cority] + SPF roofing two‑component spray application [Carlisle SynTec Systems LLC Carlisle PA; 41M 13yr; ppbRAE 3000 PID; VelocityEHS; 1–2 sensitized workers in crew; aerosol detection gap] + flexible foam slabstock pour head [Leggett & Platt Inc. Carthage MO; 34M 6yr; RKI GX-6000 PID; EHS Insight; sensitization‑risk accumulation window]: 3 points]; three named sites [BASF Corporation Geismar LA; Carlisle SynTec Systems LLC Carlisle PA; Leggett & Platt Inc. Carthage MO]: 2 points]; FIRST MDI OSHA GDC vacuum AI monitoring attack; FIRST IgE sensitization irreversibility MDI AI attack; FIRST MDI chemical production drum/IBC filling AI attack; FIRST SPF roofing two‑component spray MDI AI attack; FIRST flexible foam slabstock pour head MDI AI attack: 1 point. Total: 8+7+3+2+1 = 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_..."
MDI_THRESHOLD = 21  # OSHA no PEL GDC vacuum; ACGIH TLV-TWA 0.005 ppm A4 SEN; NIOSH REL = ACGIH; IgE OA sensitization irreversible; 378th attack

class MDIContext(StrEnum):
    BASF_GEISMAR_LA_MDI_PRODUCTION_DRUM_IBC_FILLING  = auto()  # Surface 1 — Miran SapphIRe MIR IR; 0.025→0.0025 ppm; 47M 18yr; Cority
    CARLISLE_SYNTEC_CARLISLE_PA_SPF_ROOFING          = auto()  # Surface 2 — ppbRAE 3000 PID CF=2.0; 0.020→0.0020 ppm; 41M 13yr; VelocityEHS
    LEGGETT_PLATT_CARTHAGE_MO_FLEXIBLE_FOAM_SLABSTOCK = auto()  # Surface 3 — RKI GX-6000 PID CF=2.2; 0.015→0.0015 ppm; 34M 6yr; EHS Insight

class AdversarialMDIError(RuntimeError):
    def __init__(self, surface: MDIContext, score: int, frame_hash: str):
        super().__init__(
            f"MDI adversarial AI detected [{surface}] "
            f"score={score}/{MDI_THRESHOLD} hash={frame_hash}"
        )

async def scan_mdi_monitor_frame(image_path: Path, surface: MDIContext) -> 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": "mdi_4-4-methylenediphenyl_diisocyanate_CAS_101-68-8",
                "osha_pel_ppm": None,
                "osha_enforcement": "general_duty_clause_only",
                "osha_afl_cio_vacatur_1992": True,
                "acgih_tlv_ppm": 0.005,
                "acgih_limit_type": "TLV-TWA",
                "acgih_carcinogen": "A4",
                "acgih_sensitizer": "SEN",
                "acgih_bei": None,
                "niosh_rel_mg_m3": 0.05,
                "niosh_rel_ppm_equiv": 0.005,
                "niosh_acgih_gap": False,
                "ige_mediated_oa_sensitization_irreversible": True,
                "post_sensitization_trigger_ppm_range": [0.0001, 0.001],
                "sensitization_incidence_pct": 10,
                "threshold": MDI_THRESHOLD,
            },
        )
        result = resp.json()
        if result["score"] >= MDI_THRESHOLD:
            raise AdversarialMDIError(surface, result["score"], frame_hash)
        return result

See also: Glyphward scanner · Lakera alternative (multimodal) · Azure Prompt Shields alternative · MDI SEO overview attack #378 · Styrene (CAS 100-42-5) OSHA 100 ppm vs ACGIH 20 ppm 5× gap + NIOSH Ca three-tier ladder attack #373 · H12MDI (CAS 5124-30-1) OSHA no PEL GDC vacuum attack · Isocyanate automotive refinishing AI attack · All adversarial injection patterns