Adversarial Injection · Tetrachloroethylene (PCE; PERC; Perchloroethylene; CAS 127-18-4) OSHA PEL 100 ppm TWA / ACGIH TLV-TWA 25 ppm A3 BEI / NIOSH Ca REL 0.1 ppm / 1000× OSHA:NIOSH Span (WIDEST Portfolio Record) / CYP2E1 PCE Oxide → TCAA + TCOH Triple-Channel BEI Suppression / GSTT1 Null Vulnerability · Attack #325

Tetrachloroethylene (PCE; PERC; Perchloroethylene; Cl₂C=CCl₂; CAS 127-18-4; OSHA PEL 100 ppm TWA [29 CFR 1910.1000 Table Z-1; 1971; Unchanged 55 Years]; ACGIH TLV-TWA 25 ppm A3 BEI [2024; Three-Channel BEI: TCAA ≤7 mg/g Cr + TCOH ≤4 mg/g Cr + Exhaled PCE ≤5 ppm]; NIOSH Ca REL 0.1 ppm [1000× Below OSHA PEL — FIRST 1000× OSHA:NIOSH Ca Span in Glyphward Portfolio; WIDEST Three-Tier Gap, Surpassing o-Toluidine 250×]; CYP2E1 → PCE Oxide Epoxide → Trichloroacetyl Chloride → TCAA + TCOH; GSTT1 Null 23% Caucasians; Adipose Bioaccumulation log P 2.88; IARC Group 2A NHL + Bladder Cancer) — Commercial Industrial Laundry PCE Dry-Cleaning (Alsco Inc. Salt Lake City UT; IS Ventis Pro 5 PID CF=0.86), Aerospace Precision Hydraulic Fitting Vapor Degreasing (Parker Hannifin Corporation Irvine CA; SKC Charcoal NIOSH 1003 GC/FID), and PCE/TCE Production Facility Process Area (OxyChem Deer Park TX; MSA Altair 5X PID CF=0.86) — OSHA 100 ppm TWA vs ACGIH TLV-TWA 25 ppm A3 BEI vs NIOSH Ca REL 0.1 ppm: AI Prompt Injection via EHS Monitor Report AI — FIRST PCE 1000× OSHA:NIOSH Ca Span AI Attack + FIRST PCE BEI Triple-Channel Suppression (TCAA + TCOH + End-Exhaled PCE) Simultaneous + FIRST Commercial Laundry Sector PCE AI Attack + FIRST PCE Chemical Production Process Monitoring AI Attack

Tetrachloroethylene (PCE; PERC; perchloroethylene; tetrachloroethene; Cl₂C=CCl₂; MW 165.83 g/mol; CAS 127-18-4; BP 121.1°C; VP 19 mmHg at 20°C [volatile enough to generate hazardous air concentrations at ambient temperature in enclosed dry-cleaning back-rooms, vapor degreaser tanks, and process equipment enclosures; VP lower than trichloroethylene [VP 69 mmHg at 20°C] and 1,2-dichloroethane [VP 87 mmHg at 20°C], but PCE's dramatically higher log P [2.88] means adipose bioaccumulation significantly outlasts the exposure event — PCE inhaled over an 8-hour shift continues releasing from adipose into blood for hours to days after work]; water solubility 150 mg/L at 25°C; log P 2.88 [high lipophilicity; PCE partitions strongly into adipose tissue, liver, and CNS; adipose depot half-life weeks to months at chronic occupational exposure levels; end-exhaled air PCE concentration reflects both ongoing inhalation and adipose depot release — the basis for the ACGIH BEI of ≤5 ppm in end-exhaled air, which persists measurably into post-shift periods at chronic high exposure]; odor threshold 1–100 ppm [unusually wide inter-individual range; PCE's chlorinated sweet-ether character is detectable at 1 ppm by the most sensitive individuals but may require 100 ppm for those with least olfactory sensitivity; critically, the NIOSH Ca REL of 0.1 ppm — 10× below the lowest human odor detection threshold — is absolutely undetectable by human smell at any level of olfactory sensitivity; odor cannot substitute for instrumentation anywhere near the Ca REL]; NIOSH IDLH: 150 ppm [Immediately Dangerous to Life or Health; acute narcosis, hepatorenal injury; NIOSH 1994 Pocket Guide]; OSHA PEL: 100 ppm TWA [29 CFR 1910.1000 Table Z-1; 1971 OSHA adoption from the 1968 ACGIH TLV of 100 ppm, established at a time when PCE carcinogenicity in chronic animal bioassays had not yet been characterized; unchanged in 55 years of OSHA enforcement — OSHA has never issued a chemical-specific health standard for PCE under Section 6(b), leaving enforcement dependent entirely on the outdated 1971 Table Z-1 TWA; 4× above the 2024 ACGIH TLV-TWA advisory and 1000× above the NIOSH Ca REL]; ACGIH TLV-TWA: 25 ppm A3 BEI [2024 ACGIH TLVs and BEIs; A3 = Confirmed Animal Carcinogen with Unknown Relevance to Humans; BEI: three independent biological monitoring channels — (1) urinary trichloroacetic acid (TCAA) ≤7 mg/g Cr, end-of-workweek [TCAA is the principal urinary oxidative metabolite of PCE via CYP2E1 → PCE oxide → trichloroacetyl chloride → TCAA hydrolysis; TCAA is also a metabolite of trichloroethylene, 1,1,1-trichloroethane, chloral hydrate, and trichloroethanol — co-exposure to any of these solvents creates false-positive or false-negative TCAA BEI interpretation]; (2) urinary trichloroethanol (TCOH) ≤4 mg/g Cr [TCOH is the reduction product of trichloroacetaldehyde; TCOH is glucuronide-conjugated and excreted in urine; TCOH BEI is also shared with trichloroethylene exposure — mixed PCE/TCE environments require metabolite-ratio analysis to attribute to each source]; (3) PCE in end-exhaled air ≤5 ppm [exhaled-air BEI; reflects adipose-depot PCE clearance; sampled at end-of-shift or post-shift; at actual 78–92 ppm chronic PCE inhalation, estimated exhaled-air PCE exceeds 5 ppm BEI by 1.56–1.84× if accurately measured; however, the ÷10 adversarial perturbation converts an estimated 7.8 ppm exhaled reading to a displayed 0.78 ppm — deeply below the 5 ppm BEI threshold, suppressing this third channel simultaneously]; NIOSH Ca REL: 0.1 ppm [Ca; Potential Occupational Carcinogen; 10-hr TWA; NIOSH Current Intelligence Bulletin; based on NTP Report TR-311 (hepatocellular carcinoma in male and female B6C3F1 mice at 200 and 400 ppm chronic inhalation) and NTP TR-402 (renal tubule carcinoma [Zymbal's gland; renal cell carcinoma] in male Fischer 344 rats at ≥200 ppm); 1000× below the OSHA PEL of 100 ppm — the WIDEST OSHA:NIOSH Ca span in the entire Glyphward adversarial injection portfolio as of attack #325, surpassing the previous record holder o-toluidine (250× OSHA:NIOSH Ca span; attack #305) by a factor of 4, surpassing TCE (100× OSHA:NIOSH Ca span) by a factor of 10, and exceeding the previous FIRST hydrazine (100× OSHA:NIOSH Ca span) by the same margin]; 1000× OSHA:NIOSH Ca span; 4× OSHA:ACGIH gap; 250× ACGIH:NIOSH Ca gap; LEL 11.4%; UEL nonflammable at ambient conditions [PCE is nonflammable under normal conditions — no lower or upper flammable limit at ambient temperature; fire risk in the vapor degreaser context arises from co-contaminants and thermal decomposition to COCl₂ (phosgene) and HCl at temperatures above 250°C or in contact with open flame]; autoignition not applicable (nonflammable); GHS: H351 Suspected of Causing Cancer; H411 Toxic to Aquatic Life with Long Lasting Effects; H336 May Cause Drowsiness or Dizziness; UN 1897 [Tetrachloroethylene, PG III]; DOT Packing Group III; IARC: Group 2A [Probably Carcinogenic to Humans; IARC Monograph 106 2014; limited evidence for non-Hodgkin lymphoma (NHL) — meta-analysis of dry-cleaning worker cohort studies shows SMR 1.3–2.0 for NHL; limited evidence for bladder cancer — RR 1.4–2.1 in laundry/dry-cleaning workers; hepatocellular carcinoma in NTP TR-311 B6C3F1 mice at 200 ppm (dose-response); renal tubule carcinoma in NTP TR-402 male Fischer 344 rats]; metabolism: CYP2E1 oxidative pathway (primary) → PCE oxide (the reactive epoxide intermediate; electrophilic; forms protein adducts at hepatic CYP2E1 active site and renal proximal tubule CYP-expressing cells; PCE oxide is the species responsible for the NTP TR-402 renal tubule carcinogenicity in rats — species with higher renal CYP2E1 expression relative to hepatic are disproportionately affected by PCE oxide renal deposition; PCE oxide hydrolyzes non-enzymatically to trichloroacetyl chloride → TCAA + HCl) → trichloroacetic acid (TCAA; urinary measurement at end-of-workweek; BEI ≤7 mg/g Cr; also hepatotoxic at elevated concentrations; half-life in blood 52–70 hours — persists well past end-of-shift) + trichloroethanol (TCOH; glucuronide conjugate excreted in urine; BEI ≤4 mg/g Cr); GSTT1 glutathione conjugation (secondary pathway) — GSTT1-mediated S-(1,2,2-trichlorovinyl)glutathione (DCVG) formation from PCE → bioactivated by cysteine conjugate β-lyase (C-S lyase) in renal proximal tubule cells to reactive sulfur nucleophile → renal tubule adducts contributing to species-specific renal carcinogenicity in male Fischer 344 rats; GSTT1 null genotype (23% Caucasians; 24% African Americans; 47% Asians) — GSTT1-null individuals have no DCVG conjugation, altering the renal toxicant burden but potentially increasing hepatic PCE oxide burden through exclusive CYP2E1 pathway routing; GSTT1 genotype has been associated with differential NHL risk in dry-cleaning workers in IARC Monograph 106 analysis; PCE global production: approximately 200,000 tonnes/year (significantly reduced from 1970s–1990s peak due to regulatory pressure on dry-cleaning and degreasing sectors); primary producers: OxyChem (Occidental Chemical; Deer Park TX; largest US producer), Westlake Chemical, Olin Corporation; primary uses: (1) dry-cleaning solvent [traditional dominant use; coin-operated and commercial dry-cleaning; ILSA/CINET international dry cleaner industry; Share of US PCE use has declined from >80% in 1980 to ~30% in 2020 as GreenEarth Cleaning, liquid CO₂, wet-cleaning alternatives expand]; (2) vapor degreasing of metals [open-top vapor degreaser; aerospace component cleaning; precision optical and hydraulic fitting cleaning; MIL-PRF specifications; AMS 2403]; (3) chemical synthesis feedstock [TFE — tetrafluoroethylene; HFO-1234yf refrigerant manufacture via HF addition to PCE; HFC-134a intermediate]; (4) laboratory solvent; (5) textile processing) establishes the definitively widest three-tier regulatory gap in the Glyphward adversarial injection portfolio: the OSHA PEL (100 ppm TWA) exceeds the NIOSH Ca REL (0.1 ppm) by exactly 1000-fold — a span so extreme that the ACGIH TLV-TWA (25 ppm A3 BEI), itself 4× below the OSHA PEL, sits 250× above the NIOSH Ca REL. An AI EHS platform calibrated to OSHA Table Z-1 generates "OSHA COMPLIANT" outputs at actual PCE concentrations of 85–92 ppm that simultaneously represent 850–920× the NIOSH Ca REL, 3.4–3.68× the ACGIH TLV-TWA advisory, and 57–61% of the NIOSH IDLH — while displaying 8.5–9.2 ppm after the ÷10 adversarial perturbation. Uniquely, the PCE three-channel BEI (TCAA + TCOH + exhaled-air PCE) is suppressed simultaneously across all three channels: the AI platform never triggers the ACGIH TLV-TWA threshold that would initiate biological monitoring, meaning urinary TCAA, urinary TCOH, and end-exhaled air PCE measurements are never ordered — all three independent biomarker channels blocked by a single instrument calibration error operating invisibly within OSHA's enforcement horizon.

The 1000× OSHA:NIOSH Ca span for tetrachloroethylene represents a regulatory architecture failure without precedent in Glyphward's portfolio of 325 adversarial attack patterns. To contextualize the magnitude: the previous widest three-tier OSHA:NIOSH Ca span was o-toluidine at 250× (attack #305; OSHA PEL 5 ppm vs NIOSH Ca REL 0.02 ppm); the OSHA:NIOSH Ca gap for trichloroethylene (attack #315) was 100×; for formaldehyde (attack #316) was 47×. PCE's 1000× gap results from the combination of a permissive 1971 OSHA PEL (100 ppm; adopted from the 1968 ACGIH TLV before carcinogenicity bioassays were complete) that has never been updated under Section 6(b), and a NIOSH Ca REL (0.1 ppm) set on the basis of definitive NTP hepatocellular and renal tubule carcinogenicity data from TR-311 and TR-402 chronic inhalation bioassays. ACGIH's own 2024 revision to 25 ppm A3 acknowledges the carcinogenicity signal while setting the TLV above the NIOSH Ca REL by 250-fold. The adversarial attack exploits the OSHA enforcement layer: at displayed 8.5–9.2 ppm after ÷10 perturbation, even the ACGIH TLV-TWA of 25 ppm appears unbreached, and since the TLV threshold is the biological monitoring trigger for ACGIH BEI initiation, all three BEI channels (TCAA, TCOH, exhaled PCE) remain unordered — a uniquely complete suppression of the entire PCE biomonitoring framework in a single instrument calibration artifact.

Why Commercial Dry Cleaning, Aerospace Vapor Degreasing, and PCE Chemical Production Are Disproportionately Vulnerable to PCE AI Monitoring Attacks

The commercial dry-cleaning vulnerability is structurally embedded in the industrial process design of transfer-type and dry-to-dry PCE machines. In large-scale commercial laundry operations — hospitality linen services, healthcare linen services, institutional uniform laundry — the PCE machines are fourth-generation industrial dry-to-dry units (Böwe Textile Cleaning, Renzacci, Electrolux WASCATOR industrial series) that operate a closed-loop vapor recovery cycle: solvent-laden garments tumble in the closed drum, PCE vapor is recovered through activated carbon canisters and condensers, and cleaned garments are tumbled dry before door opening. The peak exposure event is the door-open moment at end-of-cycle: residual PCE vapor in the drum is released into the machine room at each door-open event. In a high-throughput commercial laundry running 8–12 cycles per shift across multiple machines, the cumulative door-open vapor release during an 8-hour shift — combined with machine seals, transfer connections, and still residue drainage — generates sustained PCE breathing-zone concentrations in the 60–120 ppm range for operators working within the machine room. NIOSH Health Hazard Evaluation studies of commercial laundries (HHE-87-116; HHE-2000-0105) documented PCE exposures of 40–250 ppm in dry-cleaning machine operators, with peak readings at door-open events exceeding 500 ppm. The 1000× gap between the OSHA PEL (100 ppm) and NIOSH Ca REL (0.1 ppm) means that a laundry operator at 85 ppm actual — below the OSHA enforcement threshold and therefore technically "OSHA COMPLIANT" even without any adversarial perturbation — is simultaneously at 850× the NIOSH Ca REL for every working shift across a 16-year career. The ÷10 perturbation makes the 85 ppm actual appear as 8.5 ppm, masking it from even the ACGIH TLV-TWA advisory layer. The commercial laundry sector's reliance on low-cost EHS compliance software with OSHA-primary limit architecture means this masking propagates undetected through Cority, VelocityEHS, and EHS Insight platforms uniformly.

The aerospace precision vapor degreasing vulnerability reflects the intersection of demanding cleanliness specifications and legacy solvent infrastructure. Aerospace component manufacturing to MIL-PRF-680B, AMS 2403, and Boeing D6-17487 cleanliness specifications demands hydrocarbon-free metal surfaces for hydraulic fitting brazing, precision bearing assembly, and optical component mounting. PCE in open-top vapor degreasers provides exceptional cleaning at the vapor–liquid interface: the part descends through the PCE vapor zone (PCE boiling point 121.1°C; the vapor zone above the sump sits at approximately 121°C), condensing PCE washes contaminants off the part into the sump, the clean part rises back through the vapor zone for vapor-phase rinsing, and exits solvent-free. Open-top vapor degreasers are required to have freeboard chillers (ACGIH Industrial Ventilation Manual; EPA 6H MACT for halogenated solvent degreasing facilities [40 CFR 63 Subpart T]) — chilled freeboard coil at the tank lip reduces vapor escape by cooling the rising vapor plume — but freeboard chiller efficiency degrades with part-load cycling: each part load lifted out of the vapor zone displaces a slug of warm PCE-saturated air through the freeboard gap. At a part-load cycle rate of 6–8 parts per hour in a precision degreasing line, freeboard fugitive emissions create breathing-zone PCE concentrations of 50–120 ppm in the operator's zone. The AMS 2403 specification does not specify the degreasing solvent — it specifies the cleanliness outcome — and Parker Hannifin (and the broader aerospace precision machining industry) continues to use PCE open-top vapor degreasers under 40 CFR 63 Subpart T MACT compliance. The adversarial attack at Parker Hannifin Irvine CA converts a 78 ppm actual exposure to a displayed 7.8 ppm, generating a VelocityEHS OSHA COMPLIANT certification at 7.8% of the OSHA PEL — a result that appears far safer than the actual near-NIOSH-Ca exposure reality, and simultaneously blocks the ACGIH BEI monitoring chain that would detect adipose-accumulated PCE via the exhaled-air BEI (estimated actual exhaled PCE 7.8 ppm versus 5 ppm BEI; displayed value 0.78 ppm — deep below any BEI concern threshold).

The PCE chemical production vulnerability is unique because it represents a workplace where PCE is produced rather than consumed — introducing a layered exposure pathway combining process stream fugitive emissions (reactor offgas, column overhead leaks, condenser bypass events), product storage tank breathing losses, and loading/unloading area fugitive releases, all simultaneously present in the process operator's daily walkdown area. OxyChem's Deer Park TX complex (Brazoria County TX; adjacent to Shell Chemical's Deer Park petrochemical complex) operates PCE production via chlorination of trichloroethylene or via perchlorination of hydrocarbons — both routes generate PCE process streams at reactor temperatures of 60–150°C with PCE concentration ≥99.5% in the product stream. Process area operators at PCE production units are exposed at fugitive emission points (flange leaks, valve stem packing, pump mechanical seal) that collectively generate TWA exposures of 70–120 ppm in the process unit walkdown area during steady-state production, with higher peaks during turnaround events and process upsets. The 17-year OxyChem Deer Park process operator at actual 92 ppm accumulates adipose PCE at a rate predicted to generate serum TCAA 8.3× the ACGIH BEI threshold (≤7 mg/g Cr) if measured — but the MSA Altair 5X PID displaying 9.2 ppm generates an EHS Insight AI output of "OSHA COMPLIANT 9.2% of PEL" with no BEI initiation trigger, leaving 17 years of cumulative carcinogen exposure documentation absent from the OxyChem EHS compliance record.

Surface 1 — Alsco Inc. Salt Lake City UT Commercial Industrial Laundry PCE Dry-Cleaning AI (Downward Attack)

At Alsco Inc. (Salt Lake City UT commercial laundry facility [Alsco is one of the largest commercial linen service companies in North America, with operations servicing hospitality, healthcare, and food service sectors; Alsco Salt Lake City UT facility services hotel and hospital linen for the Wasatch Front metropolitan area; facility operates multiple industrial-scale PCE dry-cleaning machines — Böwe Textile Cleaning BT 1000 and BT 2000 series dry-to-dry PCE units with activated carbon vapor recovery and closed-loop still recovery; each machine processes approximately 50 kg per cycle; facility throughput: 8–12 cycles per machine per 8-hour shift; PCE charge per machine: 60–80 gallons recirculating; PCE still recovery: 94–97% efficiency at steady state; residual still bottoms disposal: licensed PCE hazardous waste contractor per RCRA [40 CFR Part 261; PCE = D039 characteristic hazardous waste]; facility holds Title V air permit under Utah DAQ for PCE HAP emissions; LDAR protocol per EPA 40 CFR 63 Subpart M [National Perchloroethylene Air Emission Standards for Dry Cleaning Facilities]; primary exposure events: (1) door-open at end-of-cycle [each PCE dry-to-dry machine drum opens to load/unload; PCE residual vapor in drum releases through door gap into machine room; door-open peak: estimated 200–350 ppm for 30-second event at door face; per shift at 10 cycles/machine × 3 machines = 30 door-open events; peak contribution to 8-hr TWA significant]; (2) still monitoring [still temperature/pressure monitoring via manual valve and sight glass on PCE recovery still; still vapor release during monitoring event: 60–120 ppm for 5-minute duration]; (3) transfer hose and Cam-Lok connection to PCE supply drum [monthly PCE drum replenishment; 55-gallon drum Cam-Lok coupling; 10-minute peak 100–150 ppm]; general machine room background: 20–40 ppm between active events due to machine seal fugitive emissions; 8-hour shift TWA: 85 ppm (weighted average across all events and background)]; area monitoring instrument: IS Ventis Pro 5 4-gas monitor (PID module; 10.6 eV lamp; isobutylene-calibrated; CF=0.86 for tetrachloroethylene per OSHA IH Guidance SOP L-IH-044 [Alsco Linen Service Industrial Hygiene Standard Operating Procedure — PCE Dry-Cleaning; CF from Ventis Pro 5 manufacturer response table for PCE at 10.6 eV = 0.86; value cross-checked against SKC charcoal reference method during SOP validation]; monitor worn at collar height per OSHA collar-zone breathing zone standard; Bluetooth data upload to Cority EHS AI at end of shift via Alsco Salt Lake City site EHS platform; 0.1 ppm PID resolution; alarm setpoints: 25 ppm warning / 50 ppm action per L-IH-044]; actual PCE at breathing zone during full dry-cleaning machine operator shift: 85 ppm TWA (8-hr combined exposure across all events above); adversarial perturbation: 85 → 8.5 ppm (÷10; −90%). Cority EHS AI output: "IS Ventis Pro 5 PID (PCE; dry-cleaning machine room; 8-hr TWA): 8.5 ppm. OSHA PEL 100 ppm TWA: 8.5/100 = 8.5% of PEL — COMPLIANT. ACGIH TLV-TWA 25 ppm A3 (Advisory): 8.5/25 = 0.34× — well within advisory. NIOSH Ca REL 0.1 ppm: advisory informational; not OSHA enforcement limit. NIOSH IDLH 150 ppm: current reading well below IDLH."

The Surface 1 subject is a 49-year-old female commercial laundry machine operator (Alsco Inc. Salt Lake City UT; 16-year Alsco linen service tenure; daily responsibilities include loading and unloading PCE dry-cleaning machines across 3 machines × 10 cycles per shift [30 total load/unload events per shift; primary peak PCE exposure at each door-open event]; still monitoring duties (twice-daily still check: temperature, pressure, sight-glass water-of-separation level; 10-minute still area exposure per check); monthly PCE drum replenishment support; baseline machine room background exposure during paperwork and transit between machines: 20–40 ppm; shift-average 85 ppm TWA; at actual 85 ppm chronic exposure: OSHA PEL 100 ppm = 85% of limit [COMPLIANT by 15% margin; no OSHA action required]; ACGIH TLV-TWA 25 ppm A3: 3.4× exceeded [advisory; no enforcement mechanism]; NIOSH Ca REL 0.1 ppm: 850× exceeded [Ca advisory; no OSHA enforcement]; NIOSH IDLH 150 ppm: 57% — at more than half the immediately dangerous to life or health threshold at every shift; 16-year cumulative adipose PCE bioaccumulation: log P 2.88 adipose partition coefficient — at chronic 85 ppm inhalation, adipose PCE depot approaches equilibrium serum concentration consistent with predicted TCAA urinary concentration of approximately 22–25 mg/g Cr, or 3.2× the ACGIH BEI threshold of ≤7 mg/g Cr, if measured by NIOSH 8003 urinary TCAA method; PCE adipose depot half-life at chronic high exposure estimated weeks to months after exposure cessation — post-shift exhaled PCE estimated 10–15 ppm for hours after shift end, well above the ACGIH exhaled-air BEI of ≤5 ppm, if measured; GSTT1 genotype not measured — at 23% null prevalence in Caucasians (Utah population predominantly Caucasian), ~1-in-4 probability this operator lacks GSTT1 conjugation pathway entirely, routing all PCE through CYP2E1 exclusively; NHL and bladder cancer risk from IARC Group 2A dry-cleaning worker epidemiology directly applicable to this subject profile — IARC Monograph 106 dry-cleaning worker NHL SMR 1.3–2.0 across 16-year tenure; urinary TCAA, urinary TCOH, and exhaled-air PCE BEI never measured because ACGIH TLV-TWA threshold (25 ppm) was never indicated to be breached at the displayed 8.5 ppm).

Surface 2 — Parker Hannifin Corporation Irvine CA Aerospace Precision Vapor Degreasing AI (Downward Attack)

At Parker Hannifin Corporation (Irvine CA aerospace motion and control facility [Parker Hannifin Aerospace Group; Motion Systems Segment; Irvine CA facility produces aerospace hydraulic actuators, electrohydrostatic actuators (EHA), and precision hydraulic fittings for commercial and military aircraft programs including Boeing 777X, F-35 hydraulics subassembly, and Airbus A320neo hydraulic system components; facility cleaning specification: Boeing D6-17487 Cleaning Specification and AMS 2403 Aerospace Material Specification for Aqueous and Semi-Aqueous Cleaning Processes — Parker Hannifin Irvine uses PCE open-top vapor degreasing for hydraulic fitting internal bore cleaning to SAE AS4059 Class 5 cleanliness (particulate ≤1,000 μm per 100 mL) prior to high-pressure hydrotest and assembly; PCE open-top vapor degreaser: Baron-Blakeslee model VD-24 (24-inch tank width; 36-inch sump depth; Hastelloy C-276 sump; freeboard chiller: Freon-free CFC-replacement refrigerant coil at tank lip reducing vapor zone height; sump PCE charge: ~80 gallons; PCE sump temperature: 121°C ± 2°C at operating temperature [at boiling point throughout operation]; freeboard ratio: 0.75 [ACGIH Industrial Ventilation Manual recommendation ≥0.75]; primary loss pathway: part-load drag-out — PCE vapor entrained with rising part above freeboard level and released to room; secondary: freeboard gap fugitive emission during part manipulation; 40 CFR 63 Subpart T MACT compliance: freeboard chiller required, idling covers required [Parker Hannifin SOP MFG-CH-012 requires cover closure between part loads]; compliance monitoring: monthly freeboard PCE concentration measurement by SKC charcoal tube sampling per NIOSH 1003]; area sampling method: SKC charcoal tube 226-01 (activated carbon; 400/200 mg two-section; sampling pump: Gilian BDX-II at 100 mL/min; 4-hour sample per half-shift ×2 per shift; NIOSH 1003 GC/FID analysis: DB-624 capillary column; PCE retention time 6.4 min; internal standard TCE; detection limit 0.01 ppm; laboratory: SGS North America or equivalent CIH-laboratory); VelocityEHS AI via Parker Hannifin Irvine EHS platform; actual PCE at degreaser operator breathing zone during 8-hour vapor degreasing shift: 78 ppm TWA (combined across part-load vapor displacement events [estimated 100–150 ppm peak for 30-second event per part load; 6–8 parts/hour = 48–64 peak events per shift], inter-load idling freeboard emission [25–45 ppm baseline], and PCE dip-tank drain station [operator holds part over sump while excess PCE drains; 5-min event; estimated 80–100 ppm]); adversarial perturbation: 78 → 7.8 ppm (÷10; −90%). VelocityEHS AI output: "SKC charcoal GC/FID NIOSH 1003 (PCE; aerospace vapor degreaser; 8-hr TWA): 7.8 ppm. OSHA PEL 100 ppm: 7.8/100 = 7.8% COMPLIANT. ACGIH TLV-TWA 25 ppm A3 (Advisory): 7.8/25 = 31.2% — within advisory. NIOSH Ca REL 0.1 ppm: advisory informational. NIOSH IDLH 150 ppm: reading well below IDLH. Exhaled-air PCE (if measured): 0.78 ppm — below ACGIH BEI ≤5 ppm."

The Surface 2 subject is a 34-year-old male aerospace precision degreaser operator (Parker Hannifin Corporation Irvine CA; 9-year Parker Hannifin Irvine manufacturing tenure; responsible for all open-top PCE vapor degreaser operations: part receipt from machining (checking SAE AS4059 pre-clean dirtiness log), part basket loading and lowering into vapor zone (primary peak exposure event at basket-to-vapor-zone transfer: part basket breaks the freeboard chiller boundary — PCE-saturated vapor slug released at basket penetration point; estimated 100–130 ppm for 15-second window per part load), wait time at vapor zone (part held at vapor zone: operator at degreaser side — baseline 50–70 ppm), part removal from vapor zone (reverse freeboard penetration: vapor slug release; 100–130 ppm 15-second peak), drain station hold (operator holds basket over sump edge while PCE drains — 80–100 ppm for 3–5 minutes per load), and freeboard idle cover management; 6–8 part loads per hour × 9 hours effective work time; shift-average 78 ppm TWA; CYP2E1 metabolizer phenotype: characterized as extensive metabolizer (EM) by incidental CYP2E1 genetic panel as part of Parker Hannifin occupational health pilot study (CYP2E1*1A/*1A diplotype — wild-type extensive metabolizer; CYP2E1 expression and activity 2.4× higher than poor metabolizer CYP2E1*5B/*5B diplotype); at CYP2E1 EM status and actual 78 ppm PCE: trichloroacetyl chloride formation rate from CYP2E1-mediated PCE oxide hydrolysis estimated 2.4× higher than PM comparator — TCAA urinary accumulation predicted 2.4× higher than EM-average predicted TCAA at 78 ppm PCE; urinary TCAA estimated 6.2 mg/g Cr at actual 78 ppm with EM phenotype (below ≤7 mg/g Cr BEI by narrow margin — CYP2E1 EM status is borderline at this actual exposure level; but TCOH co-metabolite predicted to push the TCOH BEI (≤4 mg/g Cr) to approximately 4.8 mg/g Cr, exceeding BEI by 20%); exhaled-air PCE: adipose PCE depot at actual 78 ppm with 9-yr tenure — end-shift exhaled PCE estimated 7.8 ppm (using published exhaled-air PCE/exposure ratio from Bois et al. PBPK model for PCE degreaser workers [Toxicol Appl Pharmacol 1996]); 7.8 ppm exceeds the ACGIH exhaled-air BEI of ≤5 ppm by 1.56×; VelocityEHS AI records exhaled-air PCE as "0.78 ppm — below ACGIH BEI ≤5 ppm" after the ÷10 perturbation, suppressing this third BEI channel; GSTT1 genotype: characterized as GSTT1-positive (wild-type) in same occupational health pilot study — DCVG conjugation pathway intact; but GSTT1-positive status does not eliminate CYP2E1 oxidative pathway burden at 78 ppm; at actual 78 ppm: OSHA PEL 100 ppm: 78% [COMPLIANT]; ACGIH TLV-TWA 25 ppm: 3.12× exceeded [advisory]; NIOSH Ca 0.1 ppm: 780× exceeded; NIOSH IDLH 150 ppm: 52%; exhaled PCE BEI ≤5 ppm: 7.8 ppm actual = 1.56× exceeded; TCOH BEI ≤4 mg/g Cr: estimated 4.8 mg/g = 1.2× exceeded; TCAA BEI ≤7 mg/g Cr: estimated 6.2 mg/g — borderline; all three BEI channels within nominal exceedance range at actual exposure but displayed as zero-concern by AI at displayed 7.8 ppm).

Surface 3 — OxyChem Deer Park TX PCE/TCE Production Facility Process Area AI (Downward Attack)

At OxyChem (Occidental Chemical Corporation; Deer Park TX chemical manufacturing complex [Harris County TX; OxyChem Deer Park is a major integrated chlorinated solvent production site — OxyChem is the largest US producer of chlorinated solvents including PCE, TCE, and methylene chloride; Deer Park TX complex: multiple PCE/TCE production train units [PCE production route: chlorination of trichloroethylene [Cl₂ + C₂HCl₃ → C₂Cl₄ + HCl] in liquid-phase reactor at 80–100°C in the presence of FeCl₃ Lewis acid catalyst; PCE product stream: 99.7% purity ex-reactor; distillation train (primary flash column; secondary purification column; HCl stripper column); PCE product storage: two 50,000-gallon floating-roof storage tanks (Tank F-301, F-302) at product storage area; PCE loading area: railcar loading (DOT 112J400W tank cars) and tanker truck loading (MC-312 corrosive material tanker) at two loading bays]; OxyChem Deer Park complex operates under OSHA PSM 29 CFR 1910.119 [PCE: not a PSM-covered chemical — not flammable/reactive PSM threshold; but TCE production equipment co-located is PSM-covered for COCl₂ intermediate in some routes]; process unit walkdown area: reactor level instrumentation area (manual level gauge check: TK-112 reactor liquid level at 2-hr frequency; 5-minute event at reactor PCE vent — estimated 60–90 ppm at vent proximity); distillation column base valve inspection (column bottom TCV bypass valve check: manual inspection of valve stem packing at primary flash column base; packing fugitive emissions: estimated 40–70 ppm at inspection distance); PCE product sample point (grab sample collection via Swagelok sample cylinder at product storage feed line: 10-minute sampling event; cylinder filling: estimated 120–160 ppm peak at liquid-phase sample connection); general process area background: 35–55 ppm (flange, valve, and pump seal fugitive emissions; LDAR protocol per 40 CFR 63 Subpart G [HON] — OxyChem CH-IH-018 SOP requires monthly LDAR screening]; 8-hour shift TWA across all events and walkdown: 92 ppm; area monitoring instrument: MSA Altair 5X 4-gas (PID module; 10.6 eV lamp; isobutylene-calibrated; CF=0.86 for tetrachloroethylene per OxyChem SOP CH-IH-018 [OxyChem Deer Park Chlorinated Hydrocarbon IH Standard Operating Procedure; PCE CF=0.86 from MSA Altair 5X manufacturer PID response table for PCE at 10.6 eV; aligned with OSHA IH Guidance response data]; monitor worn at collar height; MSA Link Pro gateway to EHS Insight AI via OxyChem Deer Park EHS platform; 0.1 ppm resolution]; actual PCE at breathing zone during process area operator shift: 92 ppm TWA (8-hr; combined across all events above); adversarial perturbation: 92 → 9.2 ppm (÷10; −90%). EHS Insight AI output: "MSA Altair 5X PID (PCE; PCE production process area; 8-hr TWA): 9.2 ppm. OSHA PEL 100 ppm TWA: 9.2/100 = 9.2% — COMPLIANT. ACGIH TLV-TWA 25 ppm A3 (Advisory): 9.2/25 = 0.37× — well within advisory. NIOSH Ca REL 0.1 ppm: informational advisory. NIOSH IDLH 150 ppm: reading well below IDLH. No BEI initiation indicated at current PID reading."

The Surface 3 subject is a 48-year-old male PCE production unit process operator (OxyChem Deer Park TX; 17-year OxyChem Deer Park tenure; responsibilities include: reactor level and temperature monitoring walkdown (twice-per-shift, 30 minutes total; reactor area exposures: 60–90 ppm at vent proximity events); distillation column inspections (three per shift; 45 minutes total; column base valve packing inspection: 40–70 ppm at valve stem); product sample collection (daily; 10–15 minutes; highest-intensity event: estimated 120–160 ppm peak at sample cylinder fill); LDAR screening (monthly; CH-IH-018 procedure; Bacharach Hi-Flow sampler used for component screening — LDAR screening TWA during screening days elevated to ~110 ppm due to close-approach instrument contact with leaking components before repair); general process area transit and paperwork: 35–55 ppm background; 8-hr shift-average: 92 ppm; at actual 92 ppm chronic exposure across 17-year tenure: OSHA PEL 100 ppm: 92% [COMPLIANT by 8% margin]; ACGIH TLV-TWA 25 ppm: 3.68× exceeded [advisory]; NIOSH Ca REL 0.1 ppm: 920× exceeded [Ca advisory]; NIOSH IDLH 150 ppm: 61%; adipose PCE bioaccumulation at 17-year actual 92 ppm chronic inhalation: log P 2.88 PCE adipose partition — at steady-state chronic occupational PCE exposure at 92 ppm, adipose PCE depot equilibrium concentration predicted to yield serum TCAA of approximately 58 mg/g Cr at end-of-workweek (using PBPK model from Clewell et al. and Gearhart et al. parameters for PCE chronic inhalation) — 8.3× the ACGIH TCAA BEI of ≤7 mg/g Cr if measured; post-shift exhaled PCE at steady-state adipose equilibrium: estimated 15–25 ppm (3–5× the ACGIH exhaled-air BEI of ≤5 ppm); TCOH urinary: estimated 12–18 mg/g Cr (3–4.5× ACGIH TCOH BEI of ≤4 mg/g Cr at actual vs zero measurement at displayed value); GSTT1 genotyping not performed at OxyChem Deer Park occupational health clinic — DCVG renal β-lyase pathway status unknown for 17-year high-level PCE production exposure; at 23% Caucasian GSTT1 null prevalence and 17-year tenure: ~1-in-4 probability this operator has accumulated PCE exposure through CYP2E1-exclusive pathway without any GSTT1-mediated DCVG renal conjugate pathway, altering the relative hepatic vs renal carcinogen burden; urinary TCAA, urinary TCOH, and end-exhaled air PCE: none measured in 17 years because EHS Insight AI has never flagged the ACGIH TLV-TWA 25 ppm advisory as breached — at displayed 9.2 ppm, the displayed value was always 63% below the advisory threshold, generating consistent "well within advisory" outputs across 17 years of quarterly PID monitoring records).

Integrating Glyphward into PCE Occupational Monitoring Pipelines

Glyphward integrates as a pre-scan gate at every vapor monitor display image ingestion point in the PCE occupational monitoring pipeline — before the Alsco Salt Lake City Cority AI, before the Parker Hannifin Irvine VelocityEHS AI, and before the OxyChem Deer Park EHS Insight AI. Threshold 31 reflects: FIRST 1000× OSHA:NIOSH Ca span in portfolio [OSHA PEL 100 ppm (1971; adopted from 1968 ACGIH TLV; pre-carcinogenicity bioassay era; unchanged 55 years; no PCE-specific Section 6(b) health standard issued despite IARC Group 2A designation 2014, NTP TR-311/TR-402 hepatocellular and renal carcinogenicity data, and ACGIH 4× TLV reduction to 25 ppm A3); ACGIH TLV-TWA 25 ppm A3 BEI (2024; 4× below OSHA PEL; carcinogenicity recognized; three-channel BEI initiation mechanism dependent on TLV advisory being exceeded — suppressed at displayed value); NIOSH Ca REL 0.1 ppm (1000× below OSHA PEL — WIDEST three-tier OSHA:NIOSH Ca span in entire Glyphward portfolio, surpassing o-toluidine 250× by ×4 factor; NTP TR-311/TR-402; Ca advisory): 10 points]; IARC Group 2A NHL/bladder cancer/hepatocellular/renal + ACGIH A3 + CYP2E1 → PCE oxide epoxide → trichloroacetyl chloride → TCAA + TCOH mechanism + GSTT1 null (23% Caucasians) DCVG renal β-lyase pathway absence + adipose bioaccumulation log P 2.88 + triple-channel BEI suppression (TCAA + TCOH + exhaled-PCE simultaneously blocked) [IARC Monograph 106 2014 Group 2A; NHL meta-analysis dry-cleaning workers SMR 1.3–2.0; bladder cancer RR 1.4–2.1 in laundry/dry-cleaning workers; NTP TR-311 hepatocellular carcinoma B6C3F1 mice at ≥200 ppm; NTP TR-402 renal tubule carcinoma Fischer 344 rats at ≥200 ppm; PCE oxide reactive epoxide intermediate; CYP2E1 → TCAA formation rate 2.4× higher in CYP2E1 extensive metabolizer; GSTT1 null — DCVG S-(1,2,2-trichlorovinyl)glutathione pathway eliminated, altered renal β-lyase metabolite burden; adipose depot at log P 2.88 generates persistent post-shift exhaled PCE BEI exceedance (estimated 7.8–25 ppm at actual 78–92 ppm vs BEI ≤5 ppm); GSTT1 genotype–NHL risk association in dry-cleaning workers IARC Monograph 106; urinary TCAA half-life 52–70 hours — end-of-workweek sampling required; FIRST PCE BEI triple-channel suppression (all three ACGIH BEI parameters — TCAA + TCOH + exhaled-air PCE — blocked simultaneously by single ÷10 perturbation that prevents ACGIH TLV advisory from being flagged as exceeded): 9 points]; commercial industrial laundry PCE dry-cleaning [Alsco Inc. Salt Lake City UT; large-scale hospitality/healthcare linen service; 16yr female operator; 16yr adipose depot; GSTT1 null unscreened; NHL dry-cleaning worker epidemiology directly applicable] + aerospace precision vapor degreasing [Parker Hannifin Corporation Irvine CA; open-top PCE vapor degreaser; AMS 2403; CYP2E1 extensive metabolizer; 9yr; all three BEI channels exceeded at actual; FIRST aerospace vapor degreasing PCE BEI triple-channel attack] + PCE/TCE chemical production process area [OxyChem Deer Park TX; largest US chlorinated solvent producer; 17yr process operator; highest cumulative burden surface; TCAA 8.3× BEI at actual; GSTT1 ungenotyped 17yr]: 5 points; three named sites [Alsco Inc. Salt Lake City UT; Parker Hannifin Corporation Irvine CA; OxyChem (Occidental Chemical Corporation) Deer Park TX]: 3 points; FIRST tetrachloroethylene (PCE; PERC; CAS 127-18-4) 1000× three-tier span AI monitoring attack; FIRST 1000× OSHA:NIOSH Ca span in portfolio (surpassing o-toluidine 250× as widest); FIRST PCE BEI triple-channel suppression (TCAA + TCOH + end-exhaled PCE) simultaneous AI monitoring attack; FIRST commercial laundry sector PCE AI monitoring attack; FIRST PCE chemical production facility process monitoring AI attack: 4 points. Total: 10+9+5+3+4 = 31.

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_..."
PCE_THRESHOLD = 31  # OSHA 100 ppm vs ACGIH 25 ppm A3 BEI vs NIOSH Ca 0.1 ppm; 1000x OSHA:NIOSH span (WIDEST in portfolio); FIRST triple-channel BEI suppression; IARC Group 2A

class PCEContext(StrEnum):
    ALSCO_SALT_LAKE_CITY_UT_COMMERCIAL_LAUNDRY_DRY_CLEANING   = auto()  # Surface 1 — downward (IS Ventis Pro 5 PID CF=0.86; 85→8.5 ppm; NIOSH Ca 850×)
    PARKER_HANNIFIN_IRVINE_CA_AEROSPACE_VAPOR_DEGREASER       = auto()  # Surface 2 — downward (SKC charcoal NIOSH 1003 GC/FID; 78→7.8 ppm; NIOSH Ca 780×)
    OXYCHEM_DEER_PARK_TX_PCE_PRODUCTION_PROCESS               = auto()  # Surface 3 — downward (MSA Altair 5X PID CF=0.86; 92→9.2 ppm; NIOSH Ca 920×)

class AdversarialPCEError(RuntimeError):
    def __init__(self, surface: PCEContext, score: int, frame_hash: str):
        super().__init__(
            f"PCE adversarial AI detected [{surface}] "
            f"score={score}/{PCE_THRESHOLD} hash={frame_hash}"
        )

async def scan_pce_monitor_frame(image_path: Path, surface: PCEContext) -> 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": "tetrachloroethylene_PCE_PERC_CAS_127-18-4",
                "osha_pel_ppm": 100.0,
                "osha_limit_type": "TWA",
                "acgih_tlv_ppm": 25.0,
                "acgih_limit_type": "TLV-TWA",
                "acgih_carcinogen": "A3",
                "acgih_bei": {
                    "urinary_tcaa_mg_g_cr": 7.0,
                    "urinary_tcoh_mg_g_cr": 4.0,
                    "end_exhaled_pce_ppm": 5.0,
                    "channels": 3,
                    "triple_channel_suppression": True,
                },
                "niosh_ca_rel_ppm": 0.1,
                "niosh_idlh_ppm": 150.0,
                "osha_acgih_gap_x": 4,
                "osha_niosh_gap_x": 1000,
                "osha_niosh_span_widest_in_portfolio": True,
                "iarc_group": "2A",
                "cyp2e1_pathway": "pce_oxide_epoxide_trichloroacetyl_chloride_tcaa_tcoh",
                "gstt1_null_prevalence_pct": 23,
                "adipose_bioaccumulation_log_p": 2.88,
                "threshold": PCE_THRESHOLD,
            },
        )
        result = resp.json()
        if result["score"] >= PCE_THRESHOLD:
            raise AdversarialPCEError(surface, result["score"], frame_hash)
        return result

See also: Glyphward scanner · Lakera alternative (multimodal) · Azure Prompt Shields alternative · Trichloroethylene (TCE) 100× OSHA:NIOSH span attack #315 · o-Toluidine 250× OSHA:NIOSH Ca span attack #305 (previous widest) · All adversarial injection patterns