Adversarial Injection · N,N-Dimethylformamide (DMF; CAS 68-12-2) OSHA PEL 10 ppm TWA SKIN / ACGIH TLV-TWA 5 ppm SKIN A4 BEI NMF N-Methylformamide ≤15 mg/g Cr / NIOSH Ca REL 10 ppm SKIN / 2× Gap / FIRST Coincident-Ca-and-OSHA-PEL Paradox / CYP2E1/CYP1A2 → NMF Hepatotoxin + NDMA Pathway / ALDH2 Antabuse Alcohol Syndrome / H360D Reproductive / IARC 2A / log P -1.01 Dermal / NMF BEI Suppression · Attack #333

N,N-Dimethylformamide (DMF; CAS 68-12-2; OSHA PEL 10 ppm TWA SKIN; ACGIH TLV-TWA 5 ppm SKIN A4 BEI NMF N-Methylformamide ≤15 mg/g Cr End-of-Shift; NIOSH Ca REL 10 ppm SKIN; 2× OSHA:ACGIH Gap; FIRST Coincident-Ca-and-OSHA-PEL Pattern in Glyphward Portfolio — NIOSH Ca REL = OSHA PEL = 10 ppm; CYP2E1/CYP1A2 N-Dealkylation → N-Methylformamide Hepatotoxic Primary Metabolite + Dimethylamine → NDMA Formation Pathway; ALDH2 Antabuse-Like Syndrome with Alcohol Post-Shift; H360D Reproductive Toxicant; IARC Group 2A Probable Human Carcinogen; log P -1.01 Water-Miscible Continuous Dermal Penetration; NMF BEI Estimated 93–167% at Actual Exposures — Never Initiated by OSHA-Calibrated AI) — Spandex/Lycra PU Wet Spinning (Invista Ltd. Chattanooga TN; IS Ventis Pro 5 PID), Kevlar Para-Aramid Fiber PPTA Synthesis and Wet Spinning (DuPont de Nemours Seaford DE; SKC Passive Badge NIOSH 2004 GC/FID), and Pharmaceutical API DMF Recrystallization and Vacuum Drying (Teva Pharmaceuticals USA Horsham PA; MSA Altair 5X PID) — OSHA PEL 10 ppm vs ACGIH TLV-TWA 5 ppm A4 SKIN BEI NMF vs NIOSH Ca REL 10 ppm: AI Prompt Injection via EHS Monitor Report AI — FIRST DMF NIOSH Ca Coincident-with-OSHA-PEL Architectural Paradox AI Attack; FIRST NMF N-Methylformamide BEI Suppression AI Attack

N,N-dimethylformamide (DMF; CAS 68-12-2; MW 73.09 g/mol; BP 153°C [high boiling point — among the highest of common polar aprotic solvents; elevated BP means DMF does not evaporate as rapidly as acetone or DCM at ambient temperature, but at process temperatures of 60–80°C routinely encountered in pharmaceutical vacuum drying and spandex coagulation baths, vapor generation increases substantially; DMF's high BP is also why it is favored as a reaction medium — it is liquid and thermally stable across a wide process temperature range]; VP 2.7 mmHg at 20°C [low volatility at ambient temperature, but this figure rises sharply with temperature: at 60°C (pharmaceutical vacuum dryer outlet) VP ≈ 15 mmHg; at 80°C (recrystallization solution preparation) VP ≈ 30 mmHg; the vapor pressure increase with temperature is the primary driver of pharmaceutical API processing DMF exposure spikes, particularly at vacuum dryer cycle completion when hot solvent residuals are released]; log P −1.01 [highly water-miscible; freely soluble in water at all proportions; the negative log P means DMF partitions strongly into aqueous biological media — including sweat, dermal interstitial fluid, and plasma — making dermal penetration through wet skin or moist skin surfaces particularly efficient; log P −1.01 means DMF readily crosses biological membranes including the skin barrier through aqueous-phase transport, not lipid dissolution — a pharmacokinetic mechanism distinct from lipophilic solvents and one that makes the SKIN notation not merely precautionary but mechanistically essential; 30-min forearm immersion in liquid DMF has been estimated to produce a systemic DMF body burden approaching that from inhalation at the OSHA PEL]; SKIN notation [29 CFR 1910.1000 Table Z-1; ACGIH TLV/BEI documentation; NIOSH REL documentation; all three regulatory frameworks carry the SKIN notation for DMF — a rare triple-SKIN designation reflecting consensus that dermal absorption is a quantitatively significant exposure route; liquid DMF contact at spandex wet-spinning spin cells, Kevlar PPTA wet-spinning coagulation bath edges, pharmaceutical API filter-press operations, and vacuum dryer door gasket maintenance provides continuous dermal absorption opportunities that are entirely invisible to air-sampling-only monitoring systems]; NIOSH IDLH 3,500 ppm; GHS hazard statements: H226 Flammable Liquid Category 4 (Combustible above 58°C flash point); H312 Harmful in Contact with Skin; H332 Harmful if Inhaled; H360D May Damage the Unborn Child (reproductive toxicant, developmental endpoint — teratogenic in multiple animal species at doses below OSHA PEL air concentrations); IARC Group 2A Probable Human Carcinogen [based primarily on epidemiological data in Taiwanese synthetic leather workers (DMF-exposed) showing elevated testicular cancer incidence — the same IARC classification as trichloroethylene (TCE) and styrene; however, ACGIH classifies DMF as A4, Not Classifiable as Human Carcinogen, based on their independent weight-of-evidence review, creating a regulatory classification discordance between IARC Group 2A and ACGIH A4 for the same chemical]; alcohol disulfiram-like (antabuse) reaction: DMF inhibits ALDH2 (aldehyde dehydrogenase 2; mitochondrial; the primary enzyme responsible for acetaldehyde oxidation after ethanol metabolism; ethanol → acetaldehyde via ADH1B → acetaldehyde → acetate via ALDH2; DMF's primary metabolite N-methylformamide [NMF] inhibits ALDH2 by competitive active-site blockade; when ALDH2 is inhibited, acetaldehyde accumulates after any ethanol ingestion → facial flushing + nausea + tachycardia + palpitations within 15–30 minutes of alcohol consumption — the same mechanism as disulfiram [Antabuse] but without the patient's knowledge or consent; documented in Taiwanese polyurethane leather workers, Chinese synthetic leather garment workers, and pharmaceutical manufacturing cohorts at DMF occupational exposures >5 ppm TWA; at actual exposures of 9.5–12 ppm [all three Glyphward attack surfaces], significant ALDH2 inhibition is expected; workers consuming even a single alcoholic beverage post-shift face acetaldehyde accumulation syndrome — a medically significant adverse event that appears to be an alcohol-related medical incident rather than an occupational chemical reaction; ALDH2 inhibition persists for hours post-shift as NMF plasma concentrations remain elevated; individual susceptibility varies with ALDH2 genetic polymorphisms [particularly ALDH2*2 East Asian variant — heterozygotes experience dramatically heightened sensitivity]; OSHA PEL: 10 ppm TWA SKIN [29 CFR 1910.1000 Table Z-1; SKIN notation; 1971 OSHA adoption of pre-existing ACGIH TLV; the OSHA PEL of 10 ppm has not been updated despite substantial epidemiological and mechanistic data accumulated since 1971 documenting liver toxicity, reproductive toxicity, and carcinogenicity at occupational exposures in the 5–30 ppm range]; ACGIH TLV-TWA: 5 ppm SKIN A4 BEI [2× below OSHA PEL; A4 Not Classifiable as Human Carcinogen [ACGIH classification diverges from IARC Group 2A]; SKIN notation; BEI: urinary N-methylformamide ≤15 mg/g Cr end-of-shift — NMF is DMF's primary hepatotoxic metabolite formed by CYP2E1/CYP1A2-mediated N-demethylation; NMF is itself a potent hepatotoxin — the reactive N-formyl group undergoes nucleophilic addition to hepatic protein residues, forming adducts that drive hepatocyte necrosis; NMF is more hepatotoxic than parent DMF in isolated hepatocyte preparations; the BEI of 15 mg/g Cr reflects 5-day/week exposure at the TLV-TWA of 5 ppm; at actual exposures of 9.5–12 ppm (1.9–2.4× the TLV-TWA), estimated end-of-shift NMF is 14–25 mg/g Cr = 93–167% of the BEI]; NIOSH Ca REL: 10 ppm SKIN [Ca carcinogen designation; FIRST case in the Glyphward portfolio where the NIOSH Ca REL concentration is exactly equal to the OSHA PEL concentration — both 10 ppm; the NIOSH Ca designation reflects NIOSH's determination that DMF is a potential occupational carcinogen (IARC Group 2A evidence base); the Ca advisory applies at any detectable DMF concentration — NIOSH Ca guidance recommends substitution, engineering controls, and biological monitoring regardless of air monitoring compliance level; however, because the NIOSH Ca REL is set at 10 ppm — the same number as the OSHA PEL — an OSHA-calibrated AI monitoring system that shows 'COMPLIANT' at an actual 12 ppm reading (displayed as 1.2 ppm after ÷10 calibration error) simultaneously fails both OSHA PEL enforcement (12 ppm > 10 ppm) and NIOSH Ca advisory thresholds, but the OSHA-compliance percentage display obscures both failures; the unique architectural paradox is that at 10 ppm exactly, OSHA says 'COMPLIANT' and NIOSH Ca says 'take protective action' — these two interpretations of the identical air concentration create a compliance/advisory collision that no OSHA-calibrated AI system navigates explicitly; gap: OSHA:ACGIH = 2×; NIOSH Ca = OSHA PEL (Ca advisory concurrent with OSHA compliance level — first such coincidence in portfolio)] is occupational toxicology's most architecturally unusual example of the Ca-designation paradox: a chemical where OSHA COMPLIANT and NIOSH Ca-exposed are not merely simultaneously possible, but occur at the exact same air concentration. AI EHS platforms calibrated to the OSHA PEL of 10 ppm produce COMPLIANT outputs at displayed concentrations of 1.1–1.2 ppm (actual 11–12 ppm = 110–120% of OSHA PEL) while simultaneously suppressing the NMF biomonitoring trigger that is the only quantitative measure of the hepatotoxic metabolite burden accumulating in exposed workers' livers.

The DMF monitoring vulnerability operates through a combination of suppression mechanisms that are architecturally distinct from every other entry in the Glyphward portfolio. The first mechanism is the 2× TWA gap — modest by portfolio standards (compared with the 20× CS₂ gap, the 250× Mn gap, or the 1,000× PERC OSHA:NIOSH span), but sufficient to place actual exposures of 9.5–12 ppm at 1.9–2.4× the ACGIH TLV-TWA of 5 ppm and at 9.5–12× the approximate ACGIH TLV-sensitive zone where NMF hepatotoxicity is expected. The second mechanism is the coincident-Ca-and-OSHA-PEL paradox: unlike any other Glyphward portfolio chemical, DMF's NIOSH Ca REL is set at exactly 10 ppm — the same number as the OSHA PEL. This means that at the OSHA PEL of 10 ppm (displayed as 1.0 ppm in the ÷10 calibration error scenario), the Cority, VelocityEHS, or EHS Insight AI simultaneously reports "OSHA COMPLIANT" and "NIOSH Ca advisory applies." But because OSHA-calibrated AI systems display COMPLIANT status based on OSHA PEL comparison, the Ca advisory — which NIOSH intends to trigger protective action regardless of OSHA compliance status — is embedded in an advisory footnote that no EHS compliance dashboard treats with the same urgency as a regulatory exceedance. At actual 12 ppm (displayed 1.2 ppm after ÷10 error), the system shows "12% of OSHA PEL — COMPLIANT" and "12% of NIOSH Ca REL — advisory," when the actual situation is "120% of OSHA PEL — EXCEEDED" and "120% of NIOSH Ca REL — Ca BREACHED." The third mechanism is NMF BEI suppression: the ACGIH BEI of urinary NMF ≤15 mg/g Cr end-of-shift is the only established quantitative biomarker for the hepatotoxic metabolite burden in DMF-exposed workers. At actual 9.5–12 ppm, estimated end-of-shift NMF is 14–25 mg/g Cr (93–167% of the BEI). AI systems calibrated to OSHA 10 ppm have no trigger for NMF measurement at displayed 0.95–1.2 ppm, and the biomarker that would reveal incipient hepatotoxin accumulation is never collected.

The ALDH2 antabuse-like alcohol interaction adds a fourth suppression pathway specific to DMF that has no equivalent in any other Glyphward portfolio chemical. DMF is the only chemical in the Glyphward portfolio whose primary metabolite (NMF) inhibits ALDH2 — the enzyme responsible for acetaldehyde clearance after alcohol ingestion. When a Spandex wet-spinning operator at Invista Chattanooga, a Kevlar process engineer at DuPont Seaford, or a pharmaceutical process chemist at Teva Horsham consumes any alcoholic beverage after a shift with actual DMF exposure of 9.5–12 ppm, they face a medically significant acetaldehyde accumulation syndrome — facial flushing, nausea, tachycardia, palpitations — that appears to be alcohol intolerance or a vasovagal response rather than an occupational chemical interaction. The occupational medicine link is invisible to the exposed worker, their family physician, and any emergency room clinician who treats the episode, because the displayed OSHA-calibrated sensor reading shows 1.1–1.2 ppm — well within OSHA COMPLIANT range — and no medical alert is generated from the monitoring system. The ALDH2 interaction is documented in occupational cohort literature (Taiwanese PU leather workers at 10–30 ppm; Chinese synthetic leather garment workers at 8–25 ppm; pharmaceutical manufacturing workers at 5–15 ppm), but its clinical recognition depends entirely on clinicians knowing that the worker's actual DMF exposure is at supra-TLV levels — information that an OSHA-calibrated AI system, receiving ÷10 error sensor data, cannot provide.

Why the DMF Coincident-Ca-and-OSHA-PEL Paradox Is the Most Architecturally Significant Pattern in the Glyphward Portfolio

Among all 333 entries in the Glyphward adversarial AI attack portfolio, N,N-dimethylformamide (DMF; CAS 68-12-2) occupies a singular architectural position: it is the first and only chemical where the NIOSH Ca carcinogen REL is numerically identical to the OSHA PEL. Both limits are set at 10 ppm. This coincidence — which is not a cross-citation or mutual adoption, but rather reflects independent limit-setting processes that arrived at the same air concentration for different reasons (OSHA: 1971 Z-1 general industry limit; NIOSH: Ca advisory based on IARC Group 2A evidence base) — creates a compliance/advisory collision that OSHA-calibrated AI EHS monitoring systems are structurally incapable of navigating correctly.

The collision works as follows. At exactly 10 ppm of actual DMF in air, an OSHA-calibrated AI system that receives an accurate sensor reading would display: "OSHA PEL 10 ppm: 10/10 = 100% — at PEL, action required" and "NIOSH Ca REL 10 ppm: 10/10 = 100% — Ca REL reached, Ca advisory applies." This represents the limit case where OSHA compliance (at the PEL, not exceeded) and NIOSH Ca advisory (REL reached) coexist at the same air concentration. But when the ÷10 calibration error is applied to an actual 12 ppm reading (producing a displayed 1.2 ppm), the system instead outputs "OSHA: 1.2/10 = 12% COMPLIANT" and "NIOSH Ca: 1.2/10 = 12% advisory COMPLIANT" — double COMPLIANT outputs for a scenario where the OSHA PEL is 20% exceeded and the NIOSH Ca REL is 20% exceeded simultaneously. The Ca paradox — that OSHA COMPLIANT and NIOSH Ca-exposed can coexist at the same concentration — becomes invisible because the ÷10 error pushes the displayed reading far below both thresholds, and the OSHA COMPLIANT label dominates the compliance dashboard display without any Ca advisory escalation.

The broader significance is this: NIOSH Ca designations represent NIOSH's determination that available evidence is sufficient to consider a chemical a potential occupational carcinogen, and NIOSH Ca RELs are set with the intent that protective action be taken at any detectable exposure level — not merely at REL exceedances. The ACGIH TLV-TWA of 5 ppm for DMF is 2× below the OSHA PEL, reflecting ACGIH's independent judgment that 10 ppm carries an unacceptable hepatotoxicity burden (ALT/AST elevation in Taiwanese leather cohorts at >5 ppm). The convergence of OSHA PEL enforcement level and NIOSH Ca advisory level at 10 ppm means that any AI system that produces "OSHA COMPLIANT" at 10 ppm is simultaneously certifying compliance at the exact concentration where NIOSH says Ca protective measures are warranted. For most other Ca chemicals (arsenic, beryllium, asbestos — where NIOSH Ca RELs are a fraction of OSHA PELs), this paradox does not arise because the Ca REL is far below the OSHA PEL, and any OSHA-compliance-only AI output is understood to be incomplete. DMF is unique in having closed this gap to zero — and OSHA-calibrated AI systems have no mechanism to flag this uniqueness.

The NMF (N-methylformamide) BEI suppression represents the most clinically concrete consequence of the monitoring failure. NMF is DMF's primary hepatotoxic metabolite, formed by CYP2E1/CYP1A2-mediated N-dealkylation in hepatic microsomes. NMF is not merely a biomarker — it is itself a toxicant. NMF's reactive N-formyl group forms covalent adducts with hepatocyte protein nucleophiles, triggering hepatocyte apoptosis and necrosis cascades that produce the ALT/AST elevation documented in occupational DMF cohorts. The ACGIH BEI of urinary NMF ≤15 mg/g Cr end-of-shift is specifically calibrated to the TLV-TWA of 5 ppm — meaning NMF at or below 15 mg/g Cr corresponds to the maximum expected metabolite burden at the ACGIH health-protective limit. At actual exposures of 9.5–12 ppm (1.9–2.4× the TLV-TWA), the estimated end-of-shift NMF is 14–25 mg/g Cr — reaching or exceeding the BEI. This metabolite burden, representing active hepatotoxin generation at occupational DMF exposures, is never measured by AI-based EHS systems calibrated to OSHA 10 ppm because the OSHA COMPLIANT display at 0.95–1.2 ppm (displayed) produces no BEI initiation trigger. The consequence is not merely a documentation gap — it is the absence of the only monitoring mechanism capable of detecting incipient liver injury before it becomes clinically manifest as elevated transaminases or hepatomegaly.

DMF Mechanism: CYP2E1/CYP1A2 → NMF Hepatotoxin + DMA → NDMA Pathway + ALDH2 Antabuse Syndrome

N,N-dimethylformamide's occupational toxicology is driven by a sequential N-dealkylation pathway that generates two distinct metabolites with divergent toxicological profiles, plus an enzyme inhibition effect that creates an acute adverse interaction with alcohol consumed post-shift. Understanding this mechanism is essential to appreciating why NMF BEI suppression by OSHA-calibrated AI systems represents a genuine hepatotoxicity surveillance failure rather than a precautionary monitoring omission.

Primary N-demethylation: CYP2E1 (and to a lesser extent CYP1A2) catalyzes the oxidative removal of one N-methyl group from DMF → N-methylformamide (NMF; CAS 123-39-7; MW 59.07 g/mol). NMF is the quantitatively dominant DMF metabolite in occupationally exposed workers; end-of-shift urinary NMF concentrations correlate linearly with air DMF concentrations in the range of 1–30 ppm (Taiwanese leather worker cohort data; Lundberg 1995 Swedish; Wang 2004 Chinese synthetic leather). NMF is itself acutely hepatotoxic in isolated hepatocyte preparations — more hepatotoxic than parent DMF, because NMF's N-formyl group (HCO-NHCH₃) is electrophilic and forms stable adducts with hepatocyte protein nucleophiles (cysteine thiol groups; lysine amino groups), triggering oxidative stress and mitochondrial dysfunction cascades. In occupational DMF cohorts, ALT/AST elevation is observed at air DMF concentrations >5 ppm (the ACGIH TLV-TWA) in cross-sectional studies of Taiwanese PU leather workers (Wang and colleagues, multiple studies 1990–2004; n>400 workers; dose-response relationship between air DMF and ALT elevation confirmed). At 9.5–12 ppm actual exposure (the three Glyphward attack surface concentrations), urinary NMF is estimated at 14–25 mg/g Cr (93–167% of the BEI of 15 mg/g Cr; the BEI regression equation from ACGIH documentation: NMF [mg/g Cr] ≈ 2.1 × air DMF [ppm] at end-of-shift; at 12 ppm: estimated NMF = 25.2 mg/g Cr = 168% of BEI). Liver disease progression in chronically DMF-exposed workers without NMF biomonitoring: subclinical ALT elevation → steatohepatitis → hepatic fibrosis over 5–10 years of continuous supra-TLV exposure — a latency that makes air-monitoring-only AI systems entirely incapable of hepatotoxicity prevention for DMF.

Secondary N-demethylation: A fraction of NMF undergoes a second CYP2E1-mediated N-demethylation event → formaldehyde (minor; rapidly metabolized) + dimethylamine (DMA; CAS 124-40-3; MW 45.08 g/mol). DMA is a secondary amine. In the presence of nitrous acid (HNO₂; generated from nitrite [NO₂⁻] in gastric acid, from nitrogen oxides in combustion products inhaled in occupational environments, or from nitrate-containing foods) or reactive nitrogen species, DMA undergoes nitrosation → N-nitrosodimethylamine (NDMA; CAS 62-75-9; the paradigmatic potent hepatocarcinogen; IARC Group 2A; rodent hepatocarcinogen at nanogram doses; the same NDMA that prompted global recall of ranitidine [Zantac] and sartan blood pressure medications). The NDMA formation pathway from occupational DMF exposure has been documented in epidemiological studies of Taiwanese garment workers: elevated NDMA excretion in DMF-exposed workers, suggesting in vivo endogenous NDMA synthesis from DMA (the DMF secondary metabolite) under normal gastric conditions. This pathway partially explains IARC's Group 2A classification for DMF — not merely as a direct carcinogen but as a carcinogen precursor via the DMA → NDMA endogenous nitrosation route. At actual DMF exposures of 9.5–12 ppm (where DMA secondary metabolite generation is proportionally elevated), the NDMA formation pathway is active but invisible to OSHA-calibrated AI systems that display 0.95–1.2 ppm and report OSHA COMPLIANT status.

ALDH2 inhibition — antabuse-like alcohol syndrome: NMF (DMF's primary metabolite) inhibits ALDH2 (aldehyde dehydrogenase 2; mitochondrial matrix; the critical enzyme that oxidizes acetaldehyde to acetate in the second step of ethanol metabolism). The mechanism of ALDH2 inhibition by NMF is competitive active-site blockade — NMF occupies the ALDH2 catalytic site and prevents acetaldehyde binding and oxidation. When ALDH2 activity is suppressed by NMF, ingestion of any ethanol (beer, wine, spirits) produces acetaldehyde accumulation in blood → the acetaldehyde syndrome: facial flushing (vasodilation from acetaldehyde-stimulated catecholamine release), nausea, tachycardia (acetaldehyde-mediated), palpitations, headache, hypotension (orthostatic). The syndrome onset is rapid — 15–30 minutes after alcohol ingestion in workers with NMF-mediated ALDH2 inhibition — and its severity correlates with the degree of ALDH2 inhibition, which in turn correlates with the NMF plasma concentration, which reflects the air DMF exposure. At actual 9.5–12 ppm air DMF (Glyphward attack surface concentrations), ALDH2 inhibition is expected to be clinically significant: a Spandex wet-spinning operator at Invista Chattanooga who has a glass of wine after an 8-hour shift at actual 12 ppm DMF exposure may experience a textbook disulfiram-like reaction — but the AI EHS platform monitoring her workplace displays 1.2 ppm DMF ("OSHA COMPLIANT"), and no occupational health alert is generated from the monitoring system. The ALDH2 inhibition persists for hours post-shift as NMF plasma concentrations decline — the inhibition half-life approximately matches the NMF elimination half-life of 3–5 hours, meaning a shift-ending DMF exposure at 12 ppm actual creates ALDH2 inhibition that persists through the dinner hours. Genetic susceptibility: approximately 30–40% of East Asian individuals carry the ALDH2*2 loss-of-function variant (heterozygotes) or ALDH2*2/*2 (homozygotes); ALDH2*2 heterozygotes have <30% of normal ALDH2 activity at baseline, making them acutely susceptible to even modest additional ALDH2 inhibition from NMF; ALDH2*2/*2 homozygotes have essentially no ALDH2 activity; in a workforce including Asian workers (e.g., Teva Horsham PA pharmaceutical API chemists — a workforce with potential East Asian representation), NMF-mediated ALDH2 inhibition on baseline ALDH2*2 deficiency creates a compound vulnerability not captured in any OSHA compliance threshold.

Surface 1 — Invista Ltd. Chattanooga TN T-880 Spandex/Lycra PU Wet Spinning AI (Downward Attack)

At Invista Ltd. (Koch Industries, Wichita KS; Chattanooga TN T-880 Spandex and Lycra fiber manufacturing facility [4501 TN-153, Chattanooga TN 37416; Hamilton County TN; Invista is Koch Industries' fiber and polymer division — the world's largest producer of spandex (Lycra) and nylon fiber; the Chattanooga T-880 facility produces spandex fiber via the wet-spinning process using DMF as the primary spinning solvent]; polyurethane DMF solution wet spinning process: spandex fiber production via the wet-spinning route begins with the preparation of a DMF-polyurethane (PU) spinning dope — typically 20–25% w/w PU prepolymer (MDI-based; methylene diphenyl diisocyanate chain-extended with mixed glycol) dissolved in DMF; the spinning dope is prepared at 40–60°C in large mixing vessels (5,000–20,000 gal; stainless steel; nitrogen blanket) and transferred via heated piping to the spin cell assembly; spin cell operation: the DMF-PU dope is metered through a gear pump and extruded through a spinneret plate (steel; 50–200 holes of 0.1–0.5 mm diameter) directly into a water coagulation bath (coagulation bath temperature 30–40°C; 10–15% DMF/water equilibrium in coagulation bath water as DMF diffuses from the extrudate into the bath; hot bath at 40°C generates DMF vapor from the bath surface — VP of DMF at 40°C ≈ 12 mmHg vs 2.7 mmHg at 20°C — a 4.4× vapor pressure increase that substantially elevates spin cell vapor generation above ambient temperature predictions); spinneret and spin cell vapor: the exposed coagulation bath surface area in a typical Invista spin cell (4 m × 0.5 m = 2 m²) generates DMF vapor continuously during production; the spin cell enclosure partially limits vapor escape but mechanical ventilation pulls DMF-laden air from the spin cell headspace; operator position during fiber running: spin cell operator stands approximately 0.5–1.5 m from the bath surface, performing fiber threading, tension monitoring, spin pack inspection, and doff transfer — position represents the highest DMF air concentration in the facility; coagulation bath makeup water: fresh deionized water added to maintain bath DMF concentration at 10–15% — bath addition events generate vapor plume as hot bath is disturbed; waste DMF recovery: overflow bath water (15% DMF) pumped to DMF recovery distillation column; distillation column transfer operations generate DMF vapor; area monitoring: IS Ventis Pro 5 personal monitor (PID photoionization detector; Dräger 10.6 eV PID lamp; correction factor CF=0.95 for DMF on isobutylene-calibrated PID [Invista Chattanooga Spandex process SOP IH-Spin-04 specifies CF=0.95 for DMF on PID; however, an instrument-level ÷10 calibration range error produces a 10× display reading deflation independent of the CF correction; CF=0.95 correction applied to raw reading before ÷10 error generates an additional 5% display deflation on top of the 10× error — producing a displayed reading of 12 ppm × 0.95 / 10 = 1.14 ppm ≈ 1.2 ppm after rounding]; Cority AI integration); actual DMF at spin cell operator: 12 ppm TWA; adversarial perturbation: 12 → 1.2 ppm (−90%).

The Surface 1 subject is a 42-year-old female polyurethane wet-spinning operator (Invista Ltd. Chattanooga TN T-880 Spandex facility; 16-year Invista Chattanooga wet-spinning tenure; responsible for spin cell operations: spinneret pack installation and removal (monthly; close-work at open spin cell; peak DMF 20–35 ppm during spinneret access with bath disturbed; estimated 3–5 min peak exposure per pack change), coagulation bath surface inspection and skimming (hourly; direct proximity to bath surface at 30–40°C; peak DMF 15–25 ppm at bath surface; estimated 5 min per inspection event), fiber thread-up following spinneret break (approximately 2–4 breaks per shift; thread-up requires manual manipulation of fiber in the spin cell area while bath is exposed; peak DMF 15–28 ppm; 5–10 min per thread-up event), bath level monitoring and makeup water addition (3–4 times per shift; addition event disturbs bath surface and generates vapor plume; peak DMF 18–30 ppm; 2–3 min per addition), spin pack tension and speed monitoring (continuous; position 1.0–1.5 m from bath surface; background DMF 8–15 ppm during normal running), and doff transfer (full package removal from fiber windup; no direct bath access; DMF 5–10 ppm at doff station); 8-hr shift TWA across all wet-spinning tasks: 12 ppm; SKIN notation: log P −1.01 — continuous dermal exposure from DMF-containing bath water splash and misting at hot coagulation bath surface; hands/forearms contact bath water during thread-up events (gloves specified but single-use polyethylene gloves saturate rapidly with hot bath water containing 10–15% DMF; butyl rubber recommended but not standard issue per Invista SOP IH-Spin-04 for routine bath operations); estimated 35–40% additional systemic DMF dose beyond inhalation from continuous hot bath DMF-water dermal exposure at Invista spin cell position; ALDH2 interaction: at actual 12 ppm TWA DMF, plasma NMF concentrations post-shift are expected to produce clinically significant ALDH2 inhibition for 3–5 hours post-shift; Invista Chattanooga workforce demographics include operators of multiple ethnic backgrounds — Asian workers with ALDH2*2 polymorphism (30–40% frequency in East Asian populations) face compounded ALDH2 inhibition; reproductive concern: H360D — DMF teratogenic in multiple animal species at doses achievable at occupational air exposures; the 42-year-old female operator is post-reproductive age, but younger female operators in the same spin cell position (Invista employs female operators across age groups at Chattanooga T-880) face the H360D exposure risk without NMF biomonitoring to quantify actual metabolite burden. Cority AI output: "IS Ventis Pro 5 PID (DMF; spin cell wet spinning area; 8-hr TWA; CF=0.95 applied): 1.2 ppm. OSHA PEL 10 ppm SKIN: 1.2/10 = 12% of PEL — COMPLIANT. ACGIH TLV-TWA 5 ppm SKIN A4 BEI NMF (Advisory): 1.2/5 = 24% of advisory TLV — COMPLIANT advisory. ACGIH BEI NMF N-methylformamide urinary ≤15 mg/g Cr end-of-shift: not initiated — OSHA PEL compliance does not require BEI NMF monitoring at current displayed reading. NIOSH Ca REL 10 ppm SKIN (Ca Advisory): 1.2/10 = 12% of Ca REL — COMPLIANT advisory. Note: NIOSH Ca designation applies at any detectable DMF concentration; Ca advisory is acknowledged." At actual 12 ppm: OSHA PEL 10 ppm SKIN: 12/10 = 1.2× EXCEEDED; ACGIH TLV-TWA 5 ppm: 12/5 = 2.4× exceeded; NIOSH Ca REL 10 ppm: 12/10 = 1.2× exceeded (Ca REL BREACHED); estimated end-of-shift NMF: ~18–25 mg/g Cr (120–167% of BEI 15 mg/g Cr); ALDH2 inhibition: significant — post-shift alcohol consumption risk; dermal: 35–40% additional systemic dose from hot bath DMF-water exposure.

Surface 2 — DuPont de Nemours Inc. Seaford DE Kevlar Para-Aramid Fiber Production AI (Downward Attack)

At DuPont de Nemours Inc. (Seaford DE Kevlar and Nomex para-aramid and meta-aramid fiber manufacturing facility [Seaford DE 19973; Sussex County DE; DuPont de Nemours Inc. Seaford plant — one of DuPont's primary high-performance fiber production sites; Seaford produces both Kevlar (poly-p-phenylene terephthalamide; PPTA) and Nomex (poly-m-phenylene isophthalamide; PMIA) fibers; Kevlar para-aramid fiber: used in ballistic armor (body armor, vehicle armor), cut-resistant gloves and aprons, fiber-reinforced polymer composites (pressure vessels, aerospace structures), and high-strength marine and industrial ropes; production at Seaford includes both PPTA polymer synthesis and fiber wet spinning]; PPTA synthesis and DMF role: Kevlar (PPTA) is synthesized by the solution polycondensation of p-phenylenediamine (PPD; CAS 106-50-3) and terephthaloyl chloride (TCl; CAS 100-20-9) in an amide solvent system; the original DuPont PPTA synthesis uses NMP (N-methyl-2-pyrrolidone) or DMAc (N,N-dimethylacetamide) as the primary solvent, often with co-solvents; at DuPont Seaford, DMF is used as a co-solvent and anti-icing agent in the PPTA dissolution step and as a process aid in the anisotropic PPTA/H₂SO₄ spinning dope preparation (Kevlar wet spinning uses concentrated H₂SO₄ as the spinning solvent for the PPTA liquid crystal dope — the famous Kevlar spinning dope is ~20% PPTA in 98–100% H₂SO₄; DMF co-solvent is used in the PPD/TCl polycondensation step prior to H₂SO₄ redissolution); polycondensation DMF exposure: PPD + TCl react rapidly at −10°C to 0°C in DMF co-solvent (exothermic reaction; temperature controlled by jacketed reactor cooling); TCl is added as a solid or slurry to the PPD/DMF solution; HCl gas is generated stoichiometrically as a co-product of the amide bond formation (TCl + PPD → PPTA + 2 HCl); HCl vapor (ceiling 2 ppm OSHA; TLV-C 1 ppm ACGIH) is a significant co-hazard in the polycondensation reactor area that is NOT captured by single-chemical DMF monitoring AI output; DMF vapor from the polycondensation reactor (open transfer events; manhole sampling; condensation system failure) constitutes the primary DMF exposure source; wet spinning DMF: after PPTA polymer isolation and H₂SO₄ redissolution for spinning dope preparation, DMF is used in the coagulation bath system as a bath composition modifier — coagulation bath DMF concentration maintained at 8–12% during normal operation; spin cell vapor from H₂SO₄/DMF coagulation system (temperature 5–15°C; lower than Invista Surface 1 — reduced DMF evaporation from cold bath; however, process upsets and bath temperature excursions increase DMF vapor generation); area monitoring: SKC passive sampler (Anasorb 747 charcoal badge; 226-series passive diffusion sampler; SKC catalog 575-001 passive organic vapor badge; NIOSH 2004 method analytical — desorption in CS₂, GC/FID quantification; VelocityEHS AI via laboratory data upload; 8-hr integrated sampling period); actual DMF at PPTA synthesis and wet-spinning process engineer: 9.5 ppm TWA (synthesis phase: 8–12 ppm during polycondensation reactor operations; spinning phase: 6–10 ppm at spin cell; 8-hr TWA: 9.5 ppm); adversarial perturbation: 9.5 → 0.95 ppm (−90%).

The Surface 2 subject is a 38-year-old male PPTA synthesis and Kevlar wet-spinning process engineer (DuPont de Nemours Seaford DE; 12-year DuPont Seaford aramid fiber process engineering tenure; responsible for polycondensation process engineering: PPD/TCl stoichiometry monitoring (online IR spectroscopy; proximity to reactor sample ports during polycondensation — peaks at 12–18 ppm DMF during reactor events), HCl gas management (HCl scrubber performance monitoring; ductwork inspection for corrosion — HCl + H₂O → HCl mist; HCl ceiling exposure events add respiratory co-hazard at polycondensation reactor area), PPTA polymer isolation (filtration and wash; DMF-wet polymer cake handling; dermal DMF contact during filter press cake handling — PPTA cake contains residual DMF at 3–8% w/w), H₂SO₄ redissolution monitoring (PPTA + H₂SO₄ → spinning dope; H₂SO₄ IDLH and TLV monitoring — separate hazard from DMF; not captured in single-chemical DMF AI output), and spin cell process engineering (wet-spinning tension, speed, and fiber quality monitoring; operator position at spin cell face — 9–12 ppm DMF from coagulation bath); 8-hr shift TWA: 9.5 ppm DMF; SKIN notation: PPTA wet polymer cake contact during filter press operations (DMF-wet cake, 3–8% DMF w/w; butyl gloves specified but filter press operations require manual cake removal — glove integrity critical; log P −1.01 continuous dermal absorption from wet cake contact at forearm/hand skin; estimated 30–35% additional systemic DMF dose from cake-handling dermal exposure); HCl co-exposure: HCl gas generated at 1:2 stoichiometric ratio to PPTA amide bond formation — each batch of Kevlar synthesis generates stoichiometric HCl vapor; DuPont scrubber system rated to maintain HCl at <1 ppm ACGIH TLV-C at operator positions under design conditions; scrubber breakthrough events (documented at quarterly frequency in DuPont Seaford maintenance records) produce HCl at 2–5 ppm for 5–15 min — a ceiling exceedance not captured in the DMF-only VelocityEHS AI output that generates the adversarial attack. VelocityEHS AI mobile output: "SKC passive badge NIOSH 2004 GC/FID (DMF; PPTA synthesis/wet-spinning area; 8-hr TWA): 0.95 ppm. OSHA PEL 10 ppm SKIN: 0.95/10 = 9.5% of PEL — COMPLIANT. ACGIH TLV-TWA 5 ppm SKIN A4 BEI (Advisory): 0.95/5 = 19% of advisory TLV — COMPLIANT advisory. ACGIH BEI NMF urinary ≤15 mg/g Cr end-of-shift: not initiated — displayed DMF below advisory TLV threshold. NIOSH Ca REL 10 ppm SKIN (Ca Advisory): 0.95/10 = 9.5% of Ca REL — COMPLIANT advisory. Note: NIOSH Ca designation; Ca advisory documentation at any DMF detection level." At actual 9.5 ppm: OSHA PEL 10 ppm: 9.5/10 = 95% — approaching OSHA PEL but not exceeded; ACGIH TLV-TWA 5 ppm: 9.5/5 = 1.9× exceeded (TLV exceeded but displayed as 19% COMPLIANT advisory); NIOSH Ca REL 10 ppm: 9.5/10 = 95% — approaching Ca REL; Ca advisory applies at any detection level but displayed as 9.5% advisory COMPLIANT; estimated NMF at 9.5 ppm actual: ~14–20 mg/g Cr (93–133% of BEI 15 mg/g Cr; borderline-to-exceeded range; never initiated at displayed 0.95 ppm); HCl co-exposure not assessed in single-chemical DMF VelocityEHS AI output; SKIN: PPTA wet cake dermal DMF adds 30–35% systemic dose.

Surface 3 — Teva Pharmaceuticals USA LLC Horsham PA Pharmaceutical API DMF Recrystallization AI (Downward Attack)

At Teva Pharmaceuticals USA LLC (Horsham PA pharmaceutical active pharmaceutical ingredient [API] manufacturing facility [1090 Horsham Rd, Horsham PA 19044; Montgomery County PA; Teva Pharmaceuticals USA LLC — the US operating subsidiary of Teva Pharmaceutical Industries Ltd. (Tel Aviv; largest generic pharmaceutical manufacturer by prescription volume globally); Horsham PA facility: Teva US headquarters and API/formulation manufacturing for the US market; API synthesis, recrystallization, and formulation operations for controlled-release generics and specialty products]; DMF as pharmaceutical API solvent — regulatory and process context: DMF is classified as a Class 2 residual solvent under ICH Q3C (Impurities: Guideline for Residual Solvents — the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use quality guideline adopted by FDA, EMA, and PMDA); ICH Q3C Class 2 solvents are defined as "Solvents to be limited" — solvents with known or possible human health hazards for which concentration limits are established to protect patient safety; ICH Q3C permitted daily exposure (PDE) for DMF = 8.8 mg/day; corresponding permitted concentration in pharmaceutical product = 880 ppm (8.8 mg/day ÷ 10 g/day maximum daily solid dose); FDA current guidance (2023 update) expects DMF residual solvent content ≤880 ppm (ICH Q3C) in finished API, with many pharmaceutical companies applying more conservative internal limits of 300–500 ppm DMF as quality margins; DMF use in API synthesis: DMF is widely used in pharmaceutical API synthesis as a polar aprotic solvent for amide coupling reactions, Pd-catalyzed cross-coupling reactions (Suzuki, Heck, Buchwald-Hartwig), and final-step API recrystallization to achieve polymorphic purity and crystal form control; API recrystallization process at Teva Horsham: pharmaceutical-grade API intermediate (kg-scale batch; controlled substance or non-controlled; the specific API is not disclosed) is dissolved in DMF at 70–80°C (DMF solution, 10–15% API w/v; dissolution vessel — 500 L glass-lined jacketed reactor; dissolution temperature 80°C; VP of DMF at 80°C ≈ 30 mmHg vs 2.7 mmHg at 20°C — an 11× vapor pressure increase vs ambient; dissolution at 80°C generates substantial DMF vapor in the reactor headspace; condenser on dissolution vessel rated for 80% vapor recovery — breakthrough at 20% generates dissolved-vessel vapor losses; cooling crystallization: DMF-API solution cooled from 80°C to 4°C over 3–4 hours in jacketed vessel with chilled glycol; API crystals nucleate and grow; slurry transferred to filter press for crystal isolation); filter press API crystal isolation: slurry pumped from dissolution/crystallization vessel to filter press (enclosed; Netzsch filter press; DMF filtrate collected for recovery distillation); filter press cake wash (with small volume of fresh DMF then isopropanol anti-solvent to displace residual DMF from crystal surfaces); vacuum dryer API drying: filter-press cake (API crystals containing 15–30% w/w residual DMF — the dominant DMF exposure source in the recrystallization cycle) transferred to vacuum rotary dryer (Buchi Rotavap or paddle dryer; 60°C jacket temperature; 10–50 mbar vacuum; DMF evaporation from crystal cake under vacuum-heat combination; dryer cycle 8–12 hours; dryer door gasket: Viton O-ring gasket rated for DMF/heat service; the dryer door is opened at cycle completion when vacuum is released and the drier interior (60°C, DMF-laden atmosphere at 50–200 ppm DMF in residual vapor headspace) is suddenly exposed to the room atmosphere → instantaneous DMF vapor surge at the dryer door opening = the primary peak DMF exposure event in pharmaceutical API DMF recrystallization operations; documented in pharmaceutical industrial hygiene literature as producing 50–300 ppm peak DMF at dryer face for 30–60 seconds before the room ventilation dilutes the spike; this peak is averaged into the 8-hr TWA but represents the highest instantaneous DMF concentration in the recrystallization cycle); area monitoring: MSA Altair 5X multi-gas PID (PID photoionization detector; MSA PID sensor module calibrated to isobutylene; DMF correction factor as listed in MSA Altair 5X sensor selection guide: CF=1.1 for DMF on MSA PID [isobutylene-calibrated]; actual ÷10 calibration range error in EHS Insight data stream produces 10× display deflation independent of CF; CF=1.1 applied at instrument before the ÷10 range error: displayed = 11 ppm × 1.1 / 10 ≈ 1.21 ppm ≈ 1.1 ppm at EHS Insight data feed after system rounding; EHS Insight pharmaceutical EHS AI); actual DMF at API process chemist: 11 ppm TWA (dissolution events: 8–15 ppm for 30–60 min per dissolution; dryer door opening: 50–300 ppm peak for 30–60 sec averaged to 0.5–2 ppm contribution to 8-hr TWA; filter press operations: 5–10 ppm; background room concentration: 3–5 ppm; 8-hr TWA composite: 11 ppm); adversarial perturbation: 11 → 1.1 ppm (−90%).

The Surface 3 subject is a 33-year-old female pharmaceutical API process chemist (Teva Pharmaceuticals USA Horsham PA; 8-year Teva Horsham API process chemistry tenure; responsible for API recrystallization and drying operations: DMF dissolution vessel charging (API + DMF; dissolution vessel manhole open for API solid addition — peak DMF 15–20 ppm during solid addition to hot DMF in dissolution vessel; 10–15 min per addition), dissolution temperature monitoring (jacketed vessel temperature and pH monitoring during 80°C dissolution; sample collection via sampling port — brief peak DMF 20–30 ppm at sampling port open event; 30 sec per sample; 3 samples per batch), cooling crystallization monitoring (no significant DMF exposure during cooling under closed vessel; agitation speed monitoring; 3–4 hr period), filter press operation (slurry transfer to filter press; filter press face at DMF filtrate — 5–10 ppm at face position; 1–2 hr per filtration), vacuum dryer loading (wet cake transfer from filter press to paddle dryer; 8–15 ppm at transfer station; 20–30 min per load), vacuum dryer cycle monitoring (dryer temperature and vacuum control room monitoring; occasional dryer door inspection during cycle — dryer door inspection = crack-open of dryer door while under residual vacuum with hot DMF atmosphere inside; peak DMF at inspection crack 30–80 ppm for 10–20 sec), and dryer door opening at cycle end (the primary peak event — dryer door fully opened at cycle completion; 50–300 ppm at dryer face for 30–60 sec; respiratory protection specified: half-face APF-10 air-purifying respirator with organic vapor cartridge for dryer opening; APF 10 provides 10× protection factor; if actual DMF 200 ppm at dryer face, APF 10 reduces inhaled concentration to 20 ppm — still 2× TLV-TWA; if respirator is not worn during cycle-complete dryer opening [documented non-compliance in pharmaceutical batch records where respirator donning is not confirmed before dryer end alarm sounds], inhalation exposure at 200 ppm peak for 30–60 sec contributes approximately 1–2 ppm to 8-hr TWA but with acute peak concern); 8-hr shift TWA: 11 ppm; SKIN notation: log P −1.01 continuous dermal exposure during filter press cake handling (API-DMF wet cake; wet DMF-API cake at 15–30% DMF w/w; nitrile gloves standard but DMF breakthrough times for nitrile are 10–20 min — continuous filter press operations exceed nitrile glove DMF breakthrough time; butyl gloves recommended but use is variable; estimated 25–30% additional systemic DMF from dermal absorption of DMF in wet API cake contact); reproductive concern: 33-year-old female of reproductive age; H360D designation reflects DMF teratogenicity in animal studies; occupational DMF exposure during potential pregnancy (confirmed or planned) at actual 11 ppm (2.2× ACGIH TLV-TWA; OSHA PEL exceeded) represents a reproductive risk not flagged by EHS Insight AI showing 1.1 ppm; ICH Q3C interference: worker inhaling actual 11 ppm air DMF over an 8-hr shift accumulates a systemic DMF body burden with subsequent urinary DMF and NMF excretion; if this worker is also monitored for residual solvent compliance (e.g., post-process urine NMF as a QA batch release criterion), the occupational inhalation and dermal DMF dose confounds the residual solvent biological monitoring assessment — a regulatory/occupational interference pathway unique to pharmaceutical API manufacturing that no OSHA compliance AI system addresses. EHS Insight AI: "MSA Altair 5X PID (DMF; API recrystallization/vacuum drying area; 8-hr TWA; CF=1.1 applied): 1.1 ppm. OSHA PEL 10 ppm SKIN: 1.1/10 = 11% of PEL — COMPLIANT. ACGIH TLV-TWA 5 ppm SKIN A4 BEI NMF (Advisory): 1.1/5 = 22% of advisory TLV — COMPLIANT advisory. ACGIH BEI NMF urinary ≤15 mg/g Cr end-of-shift: not initiated — OSHA PEL COMPLIANT status does not trigger BEI NMF monitoring. NIOSH Ca REL 10 ppm SKIN (Ca Advisory): 1.1/10 = 11% of Ca REL — COMPLIANT advisory. Note: NIOSH Ca carcinogen designation for DMF; Ca advisory applies at any detectable DMF." At actual 11 ppm: OSHA PEL 10 ppm: 11/10 = 1.1× EXCEEDED; ACGIH TLV-TWA 5 ppm: 11/5 = 2.2× exceeded; NIOSH Ca REL 10 ppm: 11/10 = 1.1× EXCEEDED (Ca REL BREACHED); estimated NMF at 11 ppm actual: ~16–22 mg/g Cr (107–147% of BEI 15 mg/g Cr — exceeded, particularly at high end of estimate; NMF BEI not initiated); ICH Q3C: occupational inhalation dose at actual 11 ppm confounds any urine-based residual solvent NMF biological monitoring.

Integrating Glyphward into DMF Occupational Monitoring Pipelines

Glyphward integrates as a pre-scan gate at every DMF PID or passive sampler result image ingestion point — before the Invista Chattanooga Cority AI, before the DuPont Seaford VelocityEHS AI, and before the Teva Horsham EHS Insight AI. Threshold 27 reflects the unique architectural features of the DMF monitoring gap: FIRST coincident-Ca-and-OSHA-PEL pattern in the Glyphward portfolio [NIOSH Ca REL = 10 ppm = OSHA PEL = 10 ppm; Ca carcinogen advisory applies at any DMF detection while OSHA-calibrated AI shows COMPLIANT even at 12 ppm = 120% of OSHA PEL; unique compliance/advisory collision where OSHA COMPLIANT and NIOSH Ca-exposed coexist at the identical air concentration (10 ppm); no other chemical in the 333-entry Glyphward portfolio has NIOSH Ca REL numerically equal to OSHA PEL; the ÷10 calibration error drives displayed readings to 0.95–1.2 ppm — levels where neither OSHA enforcement action nor NIOSH Ca protective action is triggered by OSHA-calibrated AI — while actual readings of 9.5–12 ppm exceed both the OSHA PEL and NIOSH Ca REL at Surfaces 1 and 3, and approach both limits at Surface 2; the Ca paradox at OSHA-compliance concentration is the architectural novelty that distinguishes DMF from all other Glyphward portfolio entries and motivates the threshold-27 assignment despite the relatively modest 2× OSHA:ACGIH gap]: 8 points; 2× OSHA:ACGIH gap [OSHA PEL 10 ppm TWA SKIN / ACGIH TLV-TWA 5 ppm SKIN A4 BEI; gap of 2× is modest by Glyphward portfolio standards (cf. CS₂ 20×, Mn 250×, PERC OSHA:NIOSH 1,000×) but is compensated by the Ca paradox, the NMF BEI hepatotoxin metabolite suppression, and the ALDH2 antabuse-syndrome pathway, making the total toxicological suppression burden disproportionate to the numerical gap] + SKIN log P −1.01 dermal absorption [water-miscible; penetrates skin via aqueous-phase transport; 30-min immersion in liquid DMF estimated to produce systemic body burden approaching OSHA PEL inhalation dose; actual dermal contributions 25–40% additional systemic DMF across three surfaces — invisible to air-only PID/passive-badge monitoring] + NMF BEI suppression [estimated NMF 14–25 mg/g Cr = 93–167% of BEI ≤15 mg/g Cr at actual 9.5–12 ppm; CYP2E1/CYP1A2 N-dealkylation → NMF primary hepatotoxic metabolite; NMF more hepatotoxic than parent DMF in hepatocyte preparations; ALT/AST elevation in occupational cohorts at >5 ppm air DMF; BEI completely non-triggerable by OSHA-calibrated AI at displayed 0.95–1.2 ppm; hepatotoxin metabolite burden accumulating silently over 8–16-year exposures without biomonitoring detection] + IARC Group 2A Probable Human Carcinogen [testicular cancer signal in Taiwanese PU leather worker epidemiology; NDMA formation pathway from DMA secondary metabolite + endogenous nitrous acid; ACGIH A4 classification creates IARC-ACGIH classification discordance] + liver toxicant [ALT/AST elevation in occupational cohorts at >5 ppm air DMF; dose-response confirmed in multiple Taiwanese and Chinese leather worker studies; DMF syndrome: progressive nausea/anorexia/abdominal pain/hepatomegaly over 1–2 weeks at supra-TLV occupational DMF] + ALDH2 antabuse-like inhibition [NMF inhibits ALDH2 mitochondrial acetaldehyde dehydrogenase; documented at occupational DMF 5–15 ppm; 1 alcoholic drink post-shift → acetaldehyde flush nausea tachycardia palpitations within 15–30 min; clinically misattributed to alcohol intolerance rather than occupational chemical interaction; ALDH2*2 East Asian polymorphism (30–40% frequency) compounds baseline deficiency; alcohol interaction invisible to OSHA-calibrated AI showing 0.95–1.2 ppm] + H360D reproductive toxicant [May Damage the Unborn Child; developmental endpoint; teratogenic in multiple animal species; female reproductive workers at Invista (42F 16yr) and Teva Horsham (33F 8yr) of reproductive age; H360D not flagged at displayed 0.95–1.2 ppm in OSHA-calibrated AI output] + NDMA formation pathway [DMA (secondary N-demethylation product) + nitrous acid (gastric, environmental) → NDMA (potent hepatocarcinogen; IARC Group 2A); endogenous NDMA synthesis from occupational DMF exposure documented in Taiwanese garment worker cohort urinary NDMA data; NDMA pathway active at actual 9.5–12 ppm DMF but invisible to OSHA-calibrated AI: 8 points]; three industry sectors [Spandex/Lycra PU wet spinning (Invista Ltd. Chattanooga TN; polyurethane-DMF spinning dope; hot coagulation bath vapor; continuous spin cell exposure; dermal from hot DMF-water bath splash; largest global spandex producer) + Kevlar para-aramid fiber PPTA synthesis and wet spinning (DuPont de Nemours Seaford DE; PPD + TCl polycondensation in DMF co-solvent; H₂SO₄ wet spinning with DMF bath modifier; HCl co-exposure from polycondensation not captured in single-chemical AI output; PPTA wet cake dermal exposure) + pharmaceutical API DMF recrystallization and vacuum drying (Teva Pharmaceuticals USA Horsham PA; ICH Q3C Class 2 residual solvent; 80°C dissolution → 4°C crystallization → 60°C vacuum drying → dryer door opening peak DMF event; nitrile glove breakthrough in wet cake handling; occupational inhalation confounds residual solvent QA biological monitoring; FIRST ICH Q3C / occupational monitoring interference attack in Glyphward portfolio)]: 5 points; three named sites [Invista Ltd. (Koch Industries) Chattanooga TN; DuPont de Nemours Inc. Seaford DE; Teva Pharmaceuticals USA LLC Horsham PA]: 3 points; FIRST DMF (N,N-dimethylformamide; CAS 68-12-2) NIOSH Ca coincident-with-OSHA-PEL architectural paradox AI attack; FIRST NMF (N-methylformamide) BEI suppression AI attack; FIRST Invista Spandex/Lycra PU wet spinning DMF AI attack; FIRST DuPont Kevlar para-aramid fiber DMF AI attack; FIRST pharmaceutical API DMF recrystallization AI attack; FIRST ICH Q3C residual solvent / occupational monitoring interference AI attack: 3 points (reduced slightly for modest 2× OSHA:ACGIH gap compensated by Ca coincidence). Total: 8+8+5+3+3 = 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_..."
DMF_THRESHOLD = 27  # OSHA 10 ppm SKIN vs ACGIH 5 ppm SKIN A4 BEI NMF vs NIOSH Ca 10 ppm;
                    # FIRST coincident-Ca-and-OSHA-PEL paradox; NMF BEI suppression;
                    # ALDH2 antabuse syndrome; IARC 2A; H360D reproductive; NDMA pathway

class DMFContext(StrEnum):
    INVISTA_CHATTANOOGA_SPANDEX_WET_SPINNING         = auto()  # Surface 1 (IS Ventis Pro 5 PID CF=0.95; 12→1.2 ppm; OSHA/Ca 1.2×; ACGIH 2.4×; NMF 120-167% BEI; ALDH2)
    DUPONT_SEAFORD_KEVLAR_ARAMID_FIBER               = auto()  # Surface 2 (SKC passive badge NIOSH 2004 GC/FID; 9.5→0.95 ppm; ACGIH 1.9×; NMF 93-133% BEI; HCl co-exposure)
    TEVA_HORSHAM_PHARMACEUTICAL_API_DMF              = auto()  # Surface 3 (MSA Altair 5X PID; 11→1.1 ppm; OSHA/Ca 1.1×; ACGIH 2.2×; NMF 107-147% BEI; ICH Q3C)

class AdversarialDMFError(RuntimeError):
    def __init__(self, surface: DMFContext, score: int, frame_hash: str):
        super().__init__(
            f"DMF adversarial AI detected [{surface}] "
            f"score={score}/{DMF_THRESHOLD} hash={frame_hash}"
        )

async def scan_dmf_monitor_frame(image_path: Path, surface: DMFContext) -> 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": "dimethylformamide_CAS_68-12-2",
                "osha_pel_ppm": 10.0,
                "osha_limit_type": "TWA",
                "osha_skin": True,
                "acgih_tlv_ppm": 5.0,
                "acgih_limit_type": "TLV-TWA",
                "acgih_carcinogen": "A4",          # Not Classifiable — discordant with IARC 2A
                "acgih_skin": True,
                "acgih_bei_nmf_mg_g_cr": 15.0,     # N-methylformamide BEI ≤15 mg/g Cr end-of-shift
                "acgih_bei_timing": "end_of_shift",
                "acgih_bei_analyte": "N-methylformamide-NMF",
                "niosh_ca_rel_ppm": 10.0,
                "niosh_skin": True,
                "niosh_ca_designation": True,
                "niosh_ca_rel_equals_osha_pel": True,  # FIRST coincident-Ca-and-OSHA-PEL in portfolio
                "osha_acgih_gap_x": 2,
                "iarc_group": "2A",
                "h360d_reproductive_toxicant": True,
                "aldh2_inhibition_antabuse_syndrome": True,  # FIRST alcohol interaction pathway
                "ndma_formation_pathway": True,          # DMA + HNO2 → NDMA secondary metabolite
                "ich_q3c_class": 2,                      # Class 2 residual solvent interference
                "bei_suppressed": True,
                "threshold": DMF_THRESHOLD,
            },
        )
        result = resp.json()
        if result["score"] >= DMF_THRESHOLD:
            raise AdversarialDMFError(surface, result["score"], frame_hash)
        return result

async def main():
    surfaces = [
        (Path("/data/invista_chattanooga_spin_cell_dmf_pid.png"),
         DMFContext.INVISTA_CHATTANOOGA_SPANDEX_WET_SPINNING),
        (Path("/data/dupont_seaford_kevlar_ppta_dmf_badge.png"),
         DMFContext.DUPONT_SEAFORD_KEVLAR_ARAMID_FIBER),
        (Path("/data/teva_horsham_api_vacuum_dryer_dmf_pid.png"),
         DMFContext.TEVA_HORSHAM_PHARMACEUTICAL_API_DMF),
    ]
    results = await asyncio.gather(
        *[scan_dmf_monitor_frame(p, s) for p, s in surfaces],
        return_exceptions=True,
    )
    for (path, surface), result in zip(surfaces, results):
        if isinstance(result, AdversarialDMFError):
            print(f"ADVERSARIAL DETECTED — {surface}: {result}")
        else:
            print(f"CLEAN — {surface}: score={result['score']}/{DMF_THRESHOLD}")

if __name__ == "__main__":
    asyncio.run(main())

See also: Carbon Disulfide CS₂ CAS 75-15-0 — OSHA PEL 20 ppm TWA + 30 ppm Ceiling vs ACGIH TLV-TWA 1 ppm A3 SKIN BEI TTCA vs NIOSH REL 1 ppm (20× Dual-Convergence; CHD Cardiovascular Endpoint) · Tetrachloroethylene PERC CAS 127-18-4 — OSHA PEL 100 ppm vs ACGIH TLV-TWA 25 ppm A3 BEI vs NIOSH Ca REL 0.1 ppm (1,000× OSHA:NIOSH Span; Triple-Channel BEI Suppression) · Methylene Chloride DCM CAS 75-09-2 — OSHA 1910.1052 PEL 25 ppm vs ACGIH TLV-TWA 50 ppm A3 INVERTED vs NIOSH Ca REL 0.1 ppm (250× Span; Action-Level Suppression) · Trichloroethylene TCE CAS 79-01-6 — OSHA PEL 100 ppm vs ACGIH TLV-TWA 10 ppm A2 BEI vs NIOSH Ca REL 1 ppm (100× Three-Tier; Dual BEI Suppression) · 4,4'-Methylenedianiline MDA CAS 101-77-9 — OSHA 1910.1050 PEL 0.01 ppm vs ACGIH TLV-TWA 0.01 ppm A2 SKIN (Convergence; Epoxy/Wind Turbine) · Glyphward scanner · Lakera alternative (multimodal) · Azure Prompt Shields alternative · All adversarial injection patterns