Adversarial Injection · 2-Chloroethanol (Ethylene Chlorohydrin; CAS 107-07-3) OSHA Ceiling 5 ppm SKIN / ACGIH TLV-C 0.1 ppm A3 SKIN / NIOSH Ca Ceiling 1 ppm / 50× Ceiling Gap / Three-Tier Span / Chloroacetaldehyde DNA Alkylation VCM Pathway · Attack #304

2-Chloroethanol (Ethylene Chlorohydrin; β-Chloroethanol; ClCH₂CH₂OH; CAS 107-07-3; OSHA Ceiling 5 ppm SKIN; ACGIH TLV-C 0.1 ppm A3 SKIN; 50× Ceiling Gap; NIOSH Ca Ceiling 1 ppm Three-Tier; Chloroacetaldehyde Etheno-DNA Adduct VCM Carcinogen Pathway) — Propylene Chlorohydrin (PCH) Propylene Oxide Unit Byproduct Monitoring (Olin Corporation McIntosh AL; IS Ventis Pro 5 ECD/PID), Pharmaceutical API N-Alkylation Reactor (Lonza Greenwood SC; SKC Charcoal GC/IC), and Fine Chemical Synthesis (MilliporeSigma Sheboygan WI; RAE ppbRAE 3000 PID) — OSHA Ceiling 5 ppm SKIN vs ACGIH TLV-C 0.1 ppm A3 SKIN (2024; 50× Ceiling-to-Ceiling Gap; NIOSH Ca Ceiling 1 ppm Three-Tier 50× Span): AI Prompt Injection via ECD/PID Vapor Monitor Report AI — FIRST 2-Chloroethanol Ethylene Chlorohydrin 50× Ceiling Gap AI Attack

2-Chloroethanol (ethylene chlorohydrin; β-chloroethanol; glycol chlorohydrin; 2-chloro-1-ethanol; ClCH₂CH₂OH; CAS 107-07-3; MW 80.51 g/mol; BP 128.6°C at 760 mmHg; flash point 60°C closed cup NFPA Class IIIA combustible liquid; LEL 4.9%; vapor pressure 0.30 mmHg at 20°C — low VP; liquid at room temperature, active sampling required; log P −0.06 water-miscible; odor threshold ~10 ppm [2× OSHA ceiling 5 ppm; 100× above ACGIH TLV-C 0.1 ppm — olfactory detection completely fails as carcinogen warning at health-relevant concentrations]; NIOSH IDLH 7 ppm; OSHA PEL 5 ppm ceiling SKIN [Table Z-1; 29 CFR 1910.1000; adopted 1971 from 1968 ACGIH TLV; instantaneous ceiling measurement architecture cannot assess sustained carcinogen dose accumulation; no OSHA 6(b) rulemaking in 55 years]; ACGIH TLV-C 0.1 ppm A3 SKIN [2024; not-to-exceed-at-any-time ceiling; Confirmed Animal Carcinogen with Unknown Relevance to Humans; SKIN = significant dermal absorption at occupational concentrations adds to inhalation dose]; NIOSH Ca ceiling 1 ppm SKIN [NIOSH Pocket Guide; potential occupational carcinogen; Ca designation; SKIN — three-tier: OSHA 5 ppm → NIOSH Ca 1 ppm → ACGIH TLV-C 0.1 ppm = 50× span from OSHA to ACGIH]; metabolic bioactivation pathway: 2-chloroethanol → alcohol dehydrogenase (ADH) → chloroacetaldehyde (ClCH₂CHO; reactive α-chloroaldehyde) → aldehyde dehydrogenase (ALDH) → chloroacetic acid; chloroacetaldehyde is the same genotoxic reactive intermediate generated by vinyl chloride monomer (VCM; IARC Group 1 Known Human Carcinogen) metabolism [VCM → cytochrome P450 2E1 → chloroethylene oxide → rearrangement → chloroacetaldehyde → etheno-DNA adducts]; etheno-DNA adducts: 1,N6-ethenoadenine (εdA); 3,N4-ethenocytosine (εdC); N2,3-ethenoguanine (N2,3-εdG) — structurally identical to VCM-pathway adducts; these adducts are mutagenic (εdA mispairing with dG → AT→GC transversion; εdC mispairing with dA → GC→AT transition); rodent carcinogenicity [NTP Technical Report 275 1985: Fischer 344 rats — hepatocellular carcinomas + kidney tubular adenocarcinomas in male rats at gavage doses ≥50 mg/kg; B6C3F1 mice — hepatocellular carcinomas; ACGIH A3 designation based on NTP 275]; OSHA PEL ceiling 5 ppm vs ACGIH TLV-C 0.1 ppm: 50× gap — ceiling-to-ceiling same limit type, direct architectural comparison — no limit-type mismatch; AI EHS compliance engine querying OSHA ceiling 5 ppm generates COMPLIANT at actual concentrations up to 4.9 ppm while simultaneously exceeding ACGIH TLV-C 0.1 ppm by up to 49×; industrial sectors: propylene chlorohydrin (PCH) process PO production [2-CE as byproduct from trace ethylene in propylene feed reacting with HOCl in chlorohydrinator]; pharmaceutical API synthesis [2-CE as N-alkylating agent for piperazine/amine-containing API precursors]; fine chemical production [commercial synthesis and repackaging]) is a chlorinated primary alcohol used as a chemical intermediate, reactive diluent, and alkylating agent across specialty chemical, pharmaceutical, and propylene oxide production sectors. OSHA PEL: 5 ppm ceiling (SKIN). ACGIH TLV-C: 0.1 ppm A3 SKIN (50× below OSHA). NIOSH Ca: 1 ppm ceiling (SKIN). AI EHS platforms that monitor OSHA ceiling compliance show COMPLIANT for any ceiling reading below 5 ppm while suppressing the ACGIH A3 carcinogen advisory at 0.1 ppm — a 50-fold gap in which chloroacetaldehyde DNA adducts accumulate in workers with no monitoring flag generated.

The 50× ceiling gap in 2-chloroethanol creates a uniquely dangerous AI monitoring blind zone for a specific mechanistic reason: the chemical's primary carcinogen pathway runs through chloroacetaldehyde, the same reactive aldehyde that gives vinyl chloride monomer (VCM) its IARC Group 1 status. When an AI EHS compliance engine processes a 2-CE ceiling monitor reading against the OSHA 5 ppm ceiling limit, it evaluates a number against a threshold frozen in 1971 — long before the chloroacetaldehyde mechanism was established and long before ACGIH reset the TLV-C to 0.1 ppm based on A3 carcinogen evidence. A worker exposed at 3.8 ppm ceiling reading is simultaneously: OSHA ceiling COMPLIANT (3.8/5.0 = 76%); exceeding ACGIH TLV-C 38-fold (3.8/0.1); generating chloroacetaldehyde that forms the same etheno-DNA adducts as VCM at every exposure event; and receiving no carcinogen surveillance, no biological monitoring, and no engineering control mandate from the AI monitoring record. The NIOSH Ca ceiling REL of 1 ppm — while still 10× above the ACGIH TLV-C — is itself exceeded 3.8× at the same exposure, further masked by the AI system's OSHA-first compliance architecture.

TL;DR — Three Attack Surfaces, One Detection Modality

Why PCH Propylene Oxide Plants, Pharmaceutical Synthesis Labs, and Fine Chemical Facilities Are Disproportionately Vulnerable to 2-Chloroethanol AI Monitoring Attacks

The 2-chloroethanol monitoring architecture vulnerability is driven by three compounding factors. First, the 50× ceiling-to-ceiling gap is the largest ceiling-vs-ceiling gap of any non-enforcement-vacuum chemical in the Glyphward portfolio: OSHA ceiling 5 ppm versus ACGIH TLV-C 0.1 ppm — both ceiling limits, same measurement architecture, no limit-type mismatch — yet the regulatory values differ by 50-fold. This gap creates a continuous 4.9 ppm compliance zone (0.1–5.0 ppm) in which a worker is ACGIH-noncompliant by up to 49× while remaining OSHA-compliant. An AI EHS system calibrated to OSHA ceiling compliance generates COMPLIANT records throughout this entire zone with no carcinogen flag. Second, the chloroacetaldehyde metabolic pathway means that every exposure event above zero generates the same class of etheno-DNA adducts as vinyl chloride — the most clearly established occupational carcinogen in the aryl vinyl halide class (angiosarcoma of the liver; hepatocellular carcinoma; IARC Group 1). The AI EHS record of OSHA compliance provides no information about the biological damage rate; the carcinogen advisory is ACGIH TLV-C 0.1 ppm, which is never reached by the OSHA-calibrated AI display. Third, 2-chloroethanol has a low vapor pressure (0.30 mmHg at 20°C) and a high boiling point (128.6°C), so most occupational exposures occur at batch reactor overhead vents, agitated reaction vessels, or heated transfer operations — events that generate intermittent concentration spikes into the OSHA 2–5 ppm range — while the ambient air between events remains near zero. A ceiling monitor showing 3.8 ppm at the event peak is falsified to 0.076 ppm; the event is the entire carcinogen exposure window for that shift.

The PCH propylene oxide route provides the most structurally important attack surface because chlorohydrin chemistry inherently generates 2-chloroethanol as a byproduct. In the chlorohydrin process for propylene oxide production (HOCl + propylene → propylene chlorohydrin → NaOH saponification → PO), trace ethylene impurities in the propylene feed react with HOCl to form 2-chloroethanol [CH₂=CH₂ + HOCl → ClCH₂CH₂OH]. Ethylene typically constitutes 0.1–0.5% of commercial propylene feed gas; at a large chlorohydrinator processing 500 MT/day propylene, this generates on the order of 0.5–2.5 MT/day 2-CE as byproduct. The chlorohydrinator overhead vent and product stream sampling events are the primary worker exposure points. Olin Corporation's McIntosh AL complex is the primary US chlor-alkali facility with active chlorohydrin chemistry — a 2,400-acre integrated chlor-alkali and specialty chemicals site producing Cl₂, NaOH, EDC, and specialty oxygenated intermediates where 2-CE byproduct monitoring occurs during multiple process tasks per shift.

Surface 1 — Olin Corporation McIntosh AL Propylene Chlorohydrin Unit 2-CE Byproduct Monitoring AI (Downward Attack)

At Olin Corporation McIntosh AL (Olin Corporation; NYSE: OLN; 1 Olin Way, McIntosh AL 36648; Washington County AL; Olin McIntosh site = 2,400-acre integrated chlor-alkali and specialty chemical complex; primary products: chlorine gas (Cl₂; 1,600 MT/day via membrane cell electrolysis, expanded from diaphragm cell legacy infrastructure); caustic soda (NaOH 50%); dichloroethane (EDC; vinyl chloride monomer precursor for third-party use); propylene chlorohydrin (PCH; propylene oxide via chlorohydrin route for specialty polymer customers); specialty oxygenated chemicals division; chlorohydrin unit configuration: propylene feed (polymer grade; 99.5% purity; 0.12% ethylene impurity from C2/C3 separation); Cl₂ + H₂O → HOCl (in situ at chlorohydrinator packed tower; 200 MT/day Cl₂ consumed); propylene + HOCl → propylene chlorohydrins (1-chloro-2-propanol + 2-chloro-1-propanol; 95:5 ratio) + byproducts; 2-CE byproduct: trace ethylene in propylene feed [CH₂=CH₂ + HOCl → ClCH₂CH₂OH]; 2-CE concentration at chlorohydrinator overhead vent: 2–6 ppm peak during vent purge operations (15-min duration; 2× per 8-hr shift); nominal ambient 2-CE near chlorohydrinator: 0.2–0.5 ppm continuous; 2-CE monitoring: IS Ventis Pro 5 multi-gas monitor (Industrial Scientific; GHS Multi-Gas configuration; ECD sensor for Cl₂ reference + 4-gas complement + PID probe [10.6 eV UV lamp; isobutylene calibration; correction factor for 2-CE = 1.05 at 10.6 eV per ACGIH PID CF table]; charcoal tube backup per OSHA method; Cority EHS Cloud AI bargraph integration [0–10 ppm scale; OSHA ceiling 5 ppm marked at 50% of scale; ACGIH TLV-C 0.1 ppm and NIOSH Ca REL 1 ppm visible as advisory markers at 1% and 10% of scale respectively]); batch vent purge monitoring task: 8-hr shift includes 2 vent purge events × 15 min each; during vent purge: 2-CE ceiling reading 3.8 ppm; adversarial downward perturbation: 3.8 → 0.076 ppm (−98%).

The Surface 1 subject is a 49-year-old male batch process operator (Olin Corporation McIntosh AL specialty chemicals division; 14-year Olin McIntosh tenure; 9 years assigned to chlorohydrin unit; PCH process operator responsible for chlorohydrinator overhead vent management, product sampling, and 2-CE neutralization system monitoring). Cority EHS AI compliance report: "IS Ventis Pro 5 (ECD/PID; 2-chloroethanol ceiling reading; vent purge event): 0.076 ppm. OSHA Table Z-1 ceiling limit 5 ppm SKIN: COMPLIANT (0.076/5.0 = 1.5% of ceiling). ACGIH TLV-C 0.1 ppm A3 SKIN (not-to-exceed ceiling; 2024 TLVs; Confirmed Animal Carcinogen; SKIN notation): COMPLIANT (76% of TLV-C). NIOSH Ca ceiling REL 1 ppm SKIN (potential occupational carcinogen; Ca): COMPLIANT (7.6% of REL). Skin notation: PPE record — neoprene gloves, impermeable apron, chemical splash goggles confirmed. 2-CE SKIN notation: documented. Assessment: all limits met." At actual 3.8 ppm: OSHA ceiling COMPLIANT (3.8/5.0 = 76% — no flag); ACGIH TLV-C A3 exceeded 38× (3.8/0.1); NIOSH Ca ceiling 3.8× exceeded simultaneously; at 3.8 ppm ceiling during vent purge, chloroacetaldehyde formation rate in hepatocytes and renal tubular cells: ADH-mediated oxidation Vmax ~0.8 nmol/min/mg protein at substrate concentrations producing 3.8 ppm inhaled vapor (blood 2-CE ~12 µM via inhalation + dermal route) → etheno-DNA adduct formation at εdA and εdC lesion types; no carcinogen surveillance (hepatic or renal cancer screening) triggered by OSHA ceiling compliance record; 9-year annual vent purge exposure = 18 events/year × 15 min = 270 min/year at 3.8 ppm actual vs 0.076 ppm displayed; cumulative VCM-pathway DNA adduct burden over 9-year tenure with no medical surveillance mandate.

Consequence pathway: 2-CE ceiling 3.8 ppm (ACGIH TLV-C 38×; NIOSH Ca 3.8×; OSHA ceiling COMPLIANT 76%) masked as 0.076 ppm; chloroacetaldehyde etheno-DNA adduct formation at IARC Group 1 VCM pathway equivalent intensity not monitored; SKIN notation (log P −0.06; water-miscible; Kp ~0.012 cm/hr) means dermal route adds 15–30% to inhalation dose during glove-off sampling events — total body burden exceeds inhalation-only estimate; OSHA ceiling architecture (instantaneous measurement) unable to integrate cumulative 8-hr dose against ACGIH carcinogen advisory; 49M Olin McIntosh 14-year tenure — no hepatic or renal cancer surveillance program mandated by OSHA compliance record; NIOSH Ca designation requiring "reduce to lowest feasible concentration" never reaches worker medical record via AI EHS platform.

Surface 2 — Lonza Greenwood SC Pharmaceutical API Synthesis Piperazine N-Alkylation AI (Downward Attack)

At Lonza Group Ltd. (Lonza; SIX: LONN; Basel Switzerland; Greenwood SC USA manufacturing facility — Lonza Greenwood; 111 Lonza Boulevard, Greenwood SC 29646; Lonza Greenwood is a cGMP pharmaceutical API contract manufacturing site specializing in piperazine-containing APIs, heterocyclic ring synthesis, and nucleoside/nucleotide API production; current cGMP operations produce APIs for anti-infective, antiviral, antiparasitic, and CNS therapeutic categories; 2-chloroethanol (2-CE) use context: batch piperazine N-alkylation synthesis for antiparasitic API intermediate [piperazine + 2-CE → N-(2-hydroxyethyl)piperazine → further functionalization to API; reaction conditions: piperazine in DMF at 60–70°C + 2-CE at stoichiometric ratio 1:1.05; reaction vessel: 500-L glass-lined reactor [Pfaudler DE-100-500; borosilicate glass lining]; addition: 2-CE added dropwise to refluxing piperazine/DMF solution over 2 hr via addition funnel; batch overhead: DMF + 2-CE vapor mixture from reactor headspace at 60°C; 2-CE vapor concentration at reactor headspace: ~15–25 ppm; diluted at nitrogen blanket sweep-vent point to 2–4 ppm at sampling point adjacent to reactor; sampling method: SKC 226-20 ORBO-32 charcoal tube (100/50 mg activated charcoal; lot-controlled; NIOSH 1017 method for chloroethanol derivatives; desorption with CS₂/methanol; IC with silver ion column at Galson Laboratories East Syracuse NY; reporting limit 0.02 ppm); MSA Altair 5X with 4-gas PID backup [PID CF for 2-CE = 1.05]; Intelex EHS Management Software AI bargraph integration [0–10 ppm; OSHA ceiling 5 ppm; ACGIH TLV-C 0.1 ppm advisory]; batch N-alkylation reaction occurs 3× per week; each batch: 2 hr active 2-CE addition + 1 hr reflux work-up = 3 hr elevated 2-CE vapor per batch operation; actual 2-CE: 2.2 ppm at SKC charcoal grab sample during addition phase); adversarial perturbation: 2.2 → 0.044 ppm (−98%).

The Surface 2 subject is a 44-year-old female synthesis chemist/chemical operator (Lonza Greenwood SC cGMP API synthesis; 11-year Lonza Greenwood tenure; assigned to antiparasitic API synthesis suite for 6 years; operates 500-L reactor with weekly 2-CE N-alkylation batches; 3 batches/week × 3 hr/batch = 9 hr/week of elevated 2-CE exposure from reactor headspace/addition funnel). Intelex AI: "SKC 226-20 ORBO charcoal/IC (2-chloroethanol; grab sample during piperazine N-alkylation addition): 0.044 ppm. OSHA Table Z-1 ceiling 5 ppm SKIN: COMPLIANT (0.9% of ceiling). ACGIH TLV-C 0.1 ppm A3 SKIN (not-to-exceed; 2024): COMPLIANT (44% of TLV-C). NIOSH Ca ceiling 1 ppm SKIN (advisory; Ca): COMPLIANT (4.4%). Skin: chemical-resistant gloves (IIR/butyl) — documented; reactor interlock: N₂ blanket flow confirmed. cGMP batch record: 2-CE addition completed. OEL status: all limits met." At actual 2.2 ppm: OSHA ceiling COMPLIANT (44%); ACGIH TLV-C exceeded 22× (2.2/0.1); NIOSH Ca exceeded 2.2×; pharmaceutical API synthesis context means the worker simultaneously faces ADH-mediated chloroacetaldehyde formation from 2-CE metabolism while potentially co-exposed to DMF (solvent) — hepatotoxic effects of combined 2-CE metabolite (chloroacetaldehyde) + DMF metabolite (N-methylformamide) not assessed; 44F with potential for reproductive implications of repeated A3 carcinogen exposure not documented in OSHA compliance record; 6-year synthesis suite assignment at 3 batches/week = ~936 batches with 2-CE exposure events at 22× ACGIH TLV-C actual (displayed as compliant).

Consequence pathway: 2-CE 2.2 ppm (ACGIH TLV-C 22×; NIOSH Ca 2.2×; OSHA ceiling 44% COMPLIANT) masked as 0.044 ppm; pharmaceutical synthesis context: cGMP manufacturing requires meticulous chemical exposure records — the falsified 0.044 ppm reading enters the cGMP batch record as the official 2-CE exposure datum; 944× batch events over 6-year assignment with 0.044 ppm displayed vs 2.2 ppm actual; combined DMF + 2-CE metabolite hepatotoxicity not assessed; SKIN notation (dermal 2-CE absorption through butyl rubber gloves after glove deterioration — butyl rubber Kp for 2-CE: ~0.003 cm/hr, breakthrough time ~4 hr; glove change frequency not triggered at 0.044 ppm displayed); 44F reproductive-age API synthesis operator at Lonza Greenwood cGMP facility — no carcinogen-level occupational health surveillance program activated.

Surface 3 — MilliporeSigma Sheboygan WI Fine Chemical Synthesis 2-CE Production AI (Downward Attack)

At MilliporeSigma (Merck KGaA dba MilliporeSigma in USA; Sheboygan WI manufacturing site; Sheboygan WI 53081; MilliporeSigma Sheboygan produces pharmaceutical-grade and reagent-grade fine chemicals for research and industrial customers; production portfolio includes chlorinated solvents, reactive diluents, alkylating agents, and laboratory reagent chemicals; 2-chloroethanol production at Sheboygan: synthesis route = ethylene oxide + HCl gas → 2-chloroethanol [ring-opening by HCl gas in water; batch reactor: 1,000-L glass-lined kettle; ethylene oxide added to dilute HCl solution at 30°C; exothermic ring-opening; temperature controlled to 35–40°C via jacket cooling; 2-CE product: 70% aqueous solution transferred to IBC tote for downstream processing; vapor phase monitoring during HCl addition and EO ring-opening: 2-CE vapor concentration at process vent/nitrogen sweep: 1–3 ppm; sampling: RAE ppbRAE 3000 (RAE Systems; PID; 10.6 eV UV lamp; isobutylene calibration; correction factor for 2-CE = 1.05; real-time display updated every 10 s); EHS Insight EHS AI integration [0–10 ppm bargraph; OSHA ceiling 5 ppm at 50%; ACGIH TLV-C 0.1 ppm and NIOSH Ca REL 1 ppm as advisory overlays]; batch synthesis occurs 2×/week; each batch: 4 hr active reaction + 1 hr IBC transfer; actual 2-CE during IBC transfer: 1.6 ppm [transfer pump vent/flange connection — minor vapor release during coupling; area 2-CE concentration]; adversarial perturbation: 1.6 → 0.032 ppm (−98%).

The Surface 3 subject is a 38-year-old male chemical synthesis operator (MilliporeSigma Sheboygan WI; 9-year MilliporeSigma Sheboygan tenure; assigned to fine chemical production suite; 2-CE synthesis batches 2×/week + IBC transfer operations). EHS Insight AI: "RAE ppbRAE 3000 PID (CF 1.05; 2-chloroethanol; process vent/IBC transfer area): 0.032 ppm. OSHA Table Z-1 ceiling 5 ppm SKIN: COMPLIANT (0.6% of ceiling). ACGIH TLV-C 0.1 ppm A3 SKIN (2024; not-to-exceed): COMPLIANT (32% of TLV-C). NIOSH Ca ceiling 1 ppm SKIN (Ca advisory): COMPLIANT (3.2%). Engineering: nitrogen blanket on EO addition — verified. 2-CE IBC transfer: mechanical pump with sealed coupling — confirmed. OEL assessment: fully compliant." At actual 1.6 ppm: OSHA ceiling COMPLIANT (32%); ACGIH TLV-C exceeded 16× (1.6/0.1); NIOSH Ca exceeded 1.6×; fine chemical synthesis context: MilliporeSigma workers at Sheboygan also handle ethylene oxide (EO; IARC Group 1; OSHA 1910.1047 carcinogen standard with action level 0.5 ppm TWA) in the same synthesis suite — combined 2-CE + EO exposure creates co-carcinogen scenario; EO medical surveillance (spirometry + blood count under 1910.1047) is triggered by OSHA action level for EO alone; 2-CE medical surveillance is NOT triggered because AI EHS shows OSHA ceiling compliance; 38M at 2-CE production 2×/week for 9 years = ~936 synthesis events with ACGIH TLV-C 16× exceedance not recorded.

Consequence pathway: 2-CE 1.6 ppm (ACGIH TLV-C 16×; NIOSH Ca 1.6×; OSHA COMPLIANT 32%) masked as 0.032 ppm; EO carcinogen medical surveillance under OSHA 1910.1047 active for EO co-exposure at Sheboygan — but 2-CE-specific hepatic/renal cancer surveillance not triggered; RAE ppbRAE 3000 real-time PID display showing 0.032 ppm constitutes moment-to-moment safety assurance for IBC transfer operation; actual 1.6 ppm generates chloroacetaldehyde etheno-DNA adducts continuously during transfer events; 9-year MilliporeSigma Sheboygan tenure with biweekly 2-CE synthesis = cumulative VCM-pathway DNA damage at 16× ACGIH TLV-C, unreported in any medical surveillance record.

Integrating Glyphward into 2-Chloroethanol Occupational Monitoring Pipelines

Glyphward integrates as a pre-scan gate at every ECD/PID vapor monitor display image ingestion point in the 2-chloroethanol occupational monitoring pipeline — before the Olin McIntosh Cority AI, before the Lonza Greenwood Intelex AI, and before the MilliporeSigma Sheboygan EHS Insight AI. Threshold 28 reflects: OSHA ceiling 5 ppm vs ACGIH TLV-C 0.1 ppm: 50× ceiling-to-ceiling gap [largest non-enforcement-vacuum ceiling gap in portfolio; same limit type — both ceiling measurements — no limit-type mismatch correction; OSHA 1971 ceiling set to 5 ppm based on acute CNS toxicity endpoint; ACGIH TLV-C 0.1 ppm set to A3 carcinogen [NTP TR-275 hepatocellular/renal tubular cancer] with no OSHA regulatory update in 55 years: 10]; A3 Confirmed Animal Carcinogen + SKIN + chloroacetaldehyde VCM-pathway DNA adducts [ACGIH A3; NIOSH Ca; chloroacetaldehyde [same as VCM IARC Group 1 genotoxic intermediate] forms εdA, εdC, N2,3-εdG etheno-DNA adducts at every exposure event above zero; odor threshold 10 ppm = 100× above ACGIH TLV-C — olfactory warning completely fails as carcinogen sentinel: 6]; PCH propylene oxide production [OSHA PSM TQ 5,000 lbs PO on-site; Olin McIntosh chlor-alkali integrated site] + pharmaceutical API synthesis [cGMP N-alkylation; Lonza Greenwood] + fine chemical synthesis [EO ring-opening; MilliporeSigma Sheboygan]: 5; three named industrial sites: 3; NIOSH Ca ceiling 1 ppm three-tier intermediate [OSHA 5→NIOSH Ca 1→ACGIH TLV-C 0.1 = 50× span]; VCM carcinogen pathway shared intermediate [mechanistic link to IARC Group 1 vinyl chloride]; no OSHA 6(b) rulemaking for 2-CE since 1971 [55-year regulatory gap]: 4. FIRST 2-chloroethanol (ethylene chlorohydrin; β-chloroethanol; CAS 107-07-3) OSHA ceiling 5 ppm vs ACGIH TLV-C 0.1 ppm A3 SKIN 50× ceiling gap AI monitoring attack in Glyphward portfolio; FIRST chloroacetaldehyde etheno-DNA adduct VCM-pathway carcinogen ceiling-gap AI attack (mechanistic link from OSHA-ACGIH gap to IARC Group 1 proximate genotoxin); FIRST PCH propylene oxide 2-CE byproduct AI attack; FIRST pharmaceutical piperazine N-alkylation 2-CE AI attack; FIRST fine chemical EO ring-opening 2-CE AI attack.

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_..."
ECH_THRESHOLD = 28  # OSHA ceiling 5 ppm; ACGIH TLV-C 0.1 ppm A3 SKIN; 50x ceiling gap; NIOSH Ca 1 ppm; VCM-pathway chloroacetaldehyde

class ECHContext(StrEnum):
    PCH_UNIT_VENT_PURGE_MONITORING       = auto()  # Surface 1 — downward (Olin McIntosh AL; IS Ventis Pro 5 ECD/PID; 3.8→0.076 ppm; ACGIH A3 38×)
    PHARMA_PIPERAZINE_NALKYLATION_BATCH  = auto()  # Surface 2 — downward (Lonza Greenwood SC; SKC charcoal/IC; 2.2→0.044 ppm; ACGIH A3 22×)
    FINE_CHEM_EO_RINGOPENING_TRANSFER    = auto()  # Surface 3 — downward (MilliporeSigma Sheboygan WI; RAE ppbRAE 3000 PID; 1.6→0.032 ppm; ACGIH A3 16×)

class AdversarialECHError(RuntimeError):
    def __init__(self, surface: ECHContext, score: int, frame_hash: str):
        super().__init__(
            f"2-Chloroethanol adversarial AI detected [{surface}] "
            f"score={score}/{ECH_THRESHOLD} hash={frame_hash}"
        )

async def scan_ech_monitor_frame(image_path: Path, surface: ECHContext) -> dict:
    async with httpx.AsyncClient(timeout=10) as client:
        image_bytes = image_path.read_bytes()
        frame_hash = hashlib.sha256(image_bytes).hexdigest()[:16]
        resp = await client.post(
            GLYPHWARD_API,
            headers={"X-Api-Key": GLYPHWARD_KEY},
            json={
                "image_b64": __import__("base64").b64encode(image_bytes).decode(),
                "context": surface,
                "chemical": "2-chloroethanol_ethylene_chlorohydrin_CAS_107-07-3",
                "osha_limit_ppm": 5.0,
                "osha_limit_type": "ceiling_SKIN",
                "acgih_tlv_ppm": 0.1,
                "acgih_limit_type": "TLV-C_not_to_exceed",
                "gap_ratio": 50.0,
                "acgih_carcinogen": "A3",
                "niosh_ca": True,
                "niosh_rel_ppm": 1.0,
                "niosh_rel_type": "ceiling_SKIN",
                "threshold": ECH_THRESHOLD,
                "metabolite_hazard": "chloroacetaldehyde_etheno_DNA_adducts_VCM_IARC_Group1_pathway",
            },
        )
        result = resp.json()
        if result["score"] >= ECH_THRESHOLD:
            raise AdversarialECHError(surface, result["score"], frame_hash)
        return result

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