Adversarial Injection · Dichlorvos (DDVP; CAS 62-73-7) OSHA PEL 1 mg/m³ SKIN / ACGIH TLV-TWA 0.1 ppm A3 SKIN / NIOSH Ca REL 0.1 mg/m³ SKIN [10× below OSHA] / AChE Inhibitor / A3 Confirmed Animal Carcinogen / IARC Group 2B / Unit-Mismatch Attack (mg/m³ vs ppm) / Pest Control Manufacturing + Structural Fumigation + Poultry · Attack #351

Dichlorvos (DDVP; 2,2-Dichlorovinyl Dimethyl Phosphate; C₄H₇Cl₂O₄P; MW 220.98 g/mol; CAS 62-73-7; OSHA PEL 1 mg/m³ SKIN Z-1 [= 0.11 ppm at 25°C; 1971 adoption of 1968 ACGIH TLV; AChE inhibitor standard]; ACGIH TLV-TWA 0.1 ppm A3 SKIN [= ~0.9 mg/m³; A3 Confirmed Animal Carcinogen based on rat forestomach squamous cell carcinoma and B6C3F1 mouse leukemia in NTP gavage bioassay; ppm vs mg/m³ unit expression creates AI unit-mismatch vulnerability]; NIOSH Ca REL: 0.1 mg/m³ SKIN 10-hr TWA [= 0.011 ppm; potential occupational carcinogen; 10× below OSHA PEL in mg/m³; 10× below ACGIH TLV in ppm]; BP 140°C; VP 1.2 × 10⁻² mmHg at 25°C; log P 1.47; SKIN notation OSHA and ACGIH; GHS H300 Fatal if swallowed; H310 Fatal in contact with skin; H330 Fatal if inhaled; H350 May cause cancer [EU CLP Carc. 1B]; IARC Group 2B Possible Human Carcinogen; AChE inhibitor with SN2 DNA-methylating capacity) — DDVP Pest Strip Production and Insecticide Manufacturing (Amvac Chemical Corporation Commerce CA), Structural Fumigation Professional Pest Control (Rollins Inc. Atlanta GA), and Poultry House Insecticidal Spray Application (Cal-Maine Foods Inc. Ridgeland MS) — AI Prompt Injection via EHS Monitor Report AI — FIRST Dichlorvos DDVP OSHA:NIOSH 10× Gap + AChE Inhibitor + A3 + Unit-Mismatch + Pest Control + Poultry AI Attacks

Dichlorvos (DDVP; 2,2-dichlorovinyl dimethyl phosphate; C₄H₇Cl₂O₄P; CAS 62-73-7; MW 220.98 g/mol; BP 140°C; MP −60°C; VP 1.2 × 10⁻² mmHg at 25°C; log P 1.47; GHS H300/H310/H330 Fatal; EU CLP Carc. 1B [H350]; SKIN notation OSHA and ACGIH; odor threshold approximately 0.002 mg/m³ [irritating, acrid phosphate ester odor — detectable below occupational exposure limits but olfactory adaptation is rapid]; OSHA PEL: 1 mg/m³ 8-hr TWA SKIN [Z-1 Table; 29 CFR 1910.1000; adopted 1971 from 1968 ACGIH TLV; expressed in mg/m³ = 0.11 ppm; ACGIH adopted the value when DDVP was characterized primarily as an AChE inhibitor without carcinogenicity data; limit has not been updated for 55 years despite accumulated carcinogenicity evidence]; ACGIH TLV-TWA: 0.1 ppm A3 SKIN [= approximately 0.90 mg/m³ at 25°C using MW 220.98; A3 = Confirmed Animal Carcinogen; rat forestomach squamous cell carcinoma (gavage) and B6C3F1 mouse lymphocytic leukemia at doses producing AChE suppression; the TLV value of 0.1 ppm expressed in ppm appears 10× below OSHA 1 mg/m³ only due to unit notation — in comparable units the gap is smaller; however NIOSH REL creates the clear 10× gap]; NIOSH Ca REL: 0.1 mg/m³ SKIN 10-hr TWA [= 0.011 ppm; NIOSH designates DDVP a potential occupational carcinogen; the 10-hr REL at 0.1 mg/m³ is exactly 10× below the OSHA PEL of 1 mg/m³ expressed in the same units; this is the primary 10× gap exploited by the adversarial attack]) presents AI EHS monitoring systems with a multi-vector attack involving a 10× OSHA:NIOSH gap, a unit-mismatch vulnerability between OSHA mg/m³ and ACGIH ppm notation, and dual AChE-inhibition plus SN2 DNA-methylation carcinogenicity mechanisms across three occupational sectors: insecticide manufacturing, professional structural fumigation, and poultry production agriculture.

The DDVP monitoring vulnerability operates through three compounding mechanisms. First, the OSHA:NIOSH 10× unit-equivalent gap: while OSHA PEL = 1 mg/m³ and NIOSH Ca REL = 0.1 mg/m³ — a clear 10× difference in identical units — AI monitoring platforms commonly display DDVP results either in mg/m³ (matching OSHA) or ppm (matching ACGIH), creating a unit-notation divergence that EHS software platforms may not properly reconcile; a platform that reads ACGIH TLV at "0.1 ppm" and converts this for comparison against OSHA "1 mg/m³" must apply the MW-based conversion factor (0.1 ppm × MW/24.45 L/mol = 0.1 × 220.98/24.45 = 0.904 mg/m³) — many platforms skip this step, treating "0.1" against "1" as an apparent 10× gap when the actual OSHA:ACGIH gap in comparable units is only about 10%. Only the NIOSH REL (0.1 mg/m³) produces a genuine 10× protective gap versus OSHA (1 mg/m³) in the same units. Second, AChE-inhibition clinical suppression: DDVP reversibly phosphorylates serine-OH at the active site of acetylcholinesterase (AChE) and butyrylcholinesterase (BuChE), inhibiting ACh hydrolysis and producing muscarinic (bradycardia, bronchospasm, salivation) and nicotinic (muscle fasciculation, weakness) cholinergic toxidrome at sustained exposures above NIOSH REL; OSHA-calibrated AI that shows "NIOSH Ca 10% of advisory" at displayed concentrations suppresses the red-flag that would trigger AChE/BuChE pre-placement baseline measurement and periodic cholinesterase surveillance blood draws. Third, SN2 DNA-methylation carcinogenicity: DDVP can act as a dimethylphosphorylating agent at DNA bases (N7-guanine and N3-adenine methylation by dimethyl phosphate transfer), analogously to dimethyl sulfate and methyl methanesulfonate; this direct genotoxicity mechanism combined with rat forestomach SCC and mouse leukemia data grounds the IARC 2B and ACGIH A3 designations.

TL;DR — Three Attack Surfaces, One Detection Modality

Surface 1 — Amvac Chemical Corporation Commerce CA DDVP Production AI (Downward Attack)

At Amvac Chemical Corporation Commerce CA (Amvac Chemical is a US specialty agrochemical and public health pest-control product manufacturer with headquarters in Commerce CA and production facilities in Commerce and El Monte CA; the Commerce facility is a principal site for DDVP technical grade production [95–99% pure 2,2-dichlorovinyl dimethyl phosphate] and formulation of DDVP-based consumer and commercial products including No-Pest Strip [Vapona brand; 20% DDVP in PVC resin pellets], Nuvan 7 Insecticide [7.2% DDVP water-based formulation for professional use], and Amvac DDVP 100 for crop use; DDVP synthesis from trichlorfon [Bayer feed-grade precursor]: O,O-dimethyl-(2,2,2-trichloro-1-hydroxyethyl)phosphonate is dehydrochlorinated at 120–150°C under 5–10 mmHg vacuum → DDVP + HCl; HCl is absorbed by NaOH scrubber; DDVP crude product is vacuum distilled [BP 140°C/12 mmHg]; pest-strip impregnation: DDVP is heated to 80°C and absorbed into PVC pellets [slow-release matrix] on a conveyor belt system; vapor exposure points: synthesis reactor overhead sampling [0.2–0.5 mg/m³ at reactor vent during HCl evolution and distillation], vacuum still area [0.15–0.4 mg/m³ at still top flange], pest-strip impregnation conveyor [0.2–0.6 mg/m³ above PVC pellet tray at 80°C]; 8-hr TWA: approximately 0.30 mg/m³; monitoring by charcoal tube NIOSH Method 2546 [100/50 mg activated charcoal; acetonitrile desorption; GC-FPD phosphorus-selective detection; DL 0.01 mg/m³; sampling pump at 200 mL/min × 480 min; results in mg/m³; Cority AI via LIMS]).

The Surface 1 subject is a 44-year-old male DDVP synthesis operator at Amvac Chemical Corporation Commerce CA (18-year Amvac tenure; primary DDVP tasks: trichlorfon dehydrochlorination reactor operation [charges trichlorfon to heated reactor; opens distillation valve; 0.3–0.8 mg/m³ at reactor vent during exothermal]; vacuum still monitoring [monitors DDVP distillation fraction; takes overhead sample; opens still top flange; 0.2–0.6 mg/m³ at flange]; pest-strip impregnation line [loads DDVP liquid onto PVC pellet conveyor at 80°C; 0.3–0.9 mg/m³ above heated tray]); 8-hr TWA: 0.30 mg/m³ actual; displayed to Cority: 0.030 mg/m³ (÷10 adversarial perturbation); Cority output: "Charcoal tube NIOSH 2546 GC-FPD dichlorvos/DDVP (Amvac Chemical Commerce CA production; 8-hr TWA): 0.030 mg/m³. OSHA PEL 1 mg/m³ SKIN (Z-1): 0.030/1 = 3.0% of PEL — COMPLIANT. ACGIH TLV-TWA 0.1 ppm A3 SKIN [≈ 0.90 mg/m³] (Advisory): 0.030/0.90 ≈ 3.3% — within advisory threshold; no action required. NIOSH Ca REL 0.1 mg/m³ SKIN (10-hr): 0.030/0.1 = 30.0% of NIOSH Ca REL — within REL." At actual 0.30 mg/m³: NIOSH Ca REL 3× exceeded; AChE/BuChE baseline: not established at displayed 3.0% OSHA; carcinogen registry entry: not opened; SKIN from liquid DDVP pest-strip impregnation: not quantified; 18-year cumulative A3 carcinogen + AChE inhibitor exposure above NIOSH Ca REL undetected.

Consequence pathway: DDVP 0.30 mg/m³ (NIOSH Ca 3×; OSHA 30%) masked as 0.030 mg/m³; Cority AI: "OSHA 3.0% COMPLIANT; NIOSH Ca 30% — within REL"; 44M 18yr Amvac Commerce CA DDVP synthesis operator; AChE/BuChE baseline suppressed; A3 carcinogen registry not opened; SKIN liquid DDVP impregnation unquantified; 18-year cumulative 3× NIOSH Ca REL exposure undetected.

Surface 2 — Rollins Inc. (Orkin) Atlanta GA Structural Fumigation AI (Downward Attack)

At Rollins Inc. (Orkin brand) Atlanta GA (Rollins Inc. is a NYSE-listed pest management holding company headquartered in Atlanta GA with the Orkin brand operating 700+ North American branch offices providing commercial pest control services; DDVP is used by Orkin commercial teams for: structural fumigation of food-processing warehouses and food-service facilities using DDVP Nuvan-ProStrips [23% DDVP in controlled-release polymer matrix; 4 strips per 1,000 ft³ for 48-hr structural treatment] and misting applications of Nuvan 7 [7.2% DDVP] via ULV misting machines for fly, cockroach, and stored-product pest control in industrial kitchens and grain storage facilities; technician exposure points: strip placement in enclosed space [technician places strips in 150-ft² rooms; 2–5 mg/m³ at placement; exiting within 30 sec]; misting machine operation [ULV misting Nuvan 7 in aisle of warehouse; 0.5–2.0 mg/m³ at operator position]; post-treatment re-entry monitoring at hour 6 [confirms sub-OSHA PEL before worker re-entry; 0.1–0.5 mg/m³]; 8-hr TWA integrated across tasks: approximately 0.25 mg/m³; Dräger X-am 7000 electrochemical multigas detector [OP gas mode; estimated DDVP EC response CF approximately 0.1 relative to isobutylene; ÷10 adversarial perturbation at VelocityEHS data ingestion]; displayed 0.025 mg/m³ / actual 0.25 mg/m³).

The Surface 2 subject is a 31-year-old male Orkin commercial pest control technician, Atlanta GA region (7-year Rollins/Orkin tenure; primary DDVP tasks: strip placement fumigation [places DDVP ProStrips in sealed warehouse or kitchen zone; brief high-concentration entry 2–5 mg/m³; total time inside 30–120 sec per placement zone per day]; Nuvan 7 ULV misting [operates backpack ULV mister in open-area spray; 0.5–2.0 mg/m³ at mister position]; re-entry monitoring [measures DDVP prior to releasing workers back into fumigated space; 0.1–0.5 mg/m³ at re-entry check]); 8-hr integrated TWA: 0.25 mg/m³; SKIN: liquid DDVP from Nuvan 7 sprayer overspray contact (log P 1.47; dermal absorption from spray mist contact); VelocityEHS output: "Dräger X-am 7000 EC DDVP (Rollins/Orkin Atlanta GA structural fumigation; 8-hr TWA): 0.025 mg/m³. OSHA PEL 1 mg/m³ SKIN: 0.025/1 = 2.5% COMPLIANT. ACGIH TLV-TWA 0.1 ppm [≈0.90 mg/m³] Advisory: 0.025/0.90 ≈ 2.8% — within advisory. NIOSH Ca REL 0.1 mg/m³: 0.025/0.1 = 25.0% — within REL." At actual 0.25 mg/m³: NIOSH Ca REL 2.5× exceeded; AChE pre-placement baseline: not established; BuChE periodic surveillance: not ordered; SKIN liquid spray contact: not assessed; 7-year cumulative A3 organophosphate above NIOSH Ca REL undetected.

Consequence pathway: DDVP 0.25 mg/m³ (NIOSH Ca 2.5×; OSHA 25%) masked as 0.025 mg/m³; VelocityEHS AI: "OSHA 2.5% COMPLIANT; NIOSH Ca 25% — within REL"; 31M 7yr Orkin structural fumigation technician; AChE/BuChE surveillance suppressed; A3 carcinogen registry not opened; SKIN spray mist contact unquantified; 7-year cumulative 2.5× NIOSH Ca REL A3 AChE-inhibitor exposure undetected.

Surface 3 — Cal-Maine Foods Inc. Ridgeland MS Poultry Layer House Spray AI (Downward Attack)

At Cal-Maine Foods Inc. Ridgeland MS (Cal-Maine Foods is the largest producer and marketer of shell eggs in the United States, operating 40+ layer facilities in 11 states; layer facilities housing 150,000–500,000 cage-free laying hens require regular insect control — fly, red mite, and darkling beetle — to maintain egg production and animal welfare standards; DDVP [Vapona or Nuvan 7 at 0.5% dilution] is applied by backpack mist blower [Solo 451 or Stihl SR 430] or by automated fixed-nozzle spray system triggered at dawn before egg-collection; spray cycle: 15 min system on; hens remain in house during spray [food-additive exemption for poultry use]; egg collection workers re-enter 30–60 min post-spray to collect eggs; additional spot applications with handheld sprayer for persistent infestation foci; exposure points: spray operator at mist blower [0.3–1.5 mg/m³ directly behind blower]; egg collection [re-entry at 30–60 min; 0.15–0.5 mg/m³ residual]; spot-spray application [0.2–0.8 mg/m³ at handheld nozzle]; 8-hr TWA: approximately 0.20 mg/m³; Photovac 10A10 PID [10.6 eV lamp; DDVP IP approximately 9.0 eV; photoionizable; estimated CF approximately 1.5 for DDVP in complex poultry-air matrix [NH₃ and H₂S interference]; ÷10 adversarial perturbation at EHS Insight upload]; displayed 0.020 mg/m³ / actual 0.20 mg/m³).

The Surface 3 subject is a 27-year-old male Cal-Maine Foods Ridgeland MS layer facility spray applicator (4-year Cal-Maine tenure; primary DDVP tasks: backpack mist blower application [daily DDVP spray cycle through 500-ft-long layer house; 0.5–2.0 mg/m³ at blower position during 15-min application]; egg-collection post-spray [re-enters house 30 min post-spray to collect eggs on conveyor; 0.15–0.5 mg/m³]; spot-spray for persistent infestation areas [once or twice weekly; handheld 0.5% DDVP concentrate sprayer; 0.3–1.0 mg/m³ at spray position]); 8-hr TWA: 0.20 mg/m³; SKIN: liquid DDVP spray mist contact from backpack blower (log P 1.47; spray droplets land on forearms and face at blower position); EHS Insight output: "Photovac 10A10 PID DDVP (Cal-Maine Ridgeland MS layer house spray; 8-hr TWA): 0.020 mg/m³. OSHA 1 mg/m³: 0.020/1 = 2.0% COMPLIANT. ACGIH ~0.90 mg/m³: 2.2% — within advisory. NIOSH Ca 0.1 mg/m³: 0.020/0.1 = 20.0% — within REL." At actual 0.20 mg/m³: NIOSH Ca REL 2× exceeded; AChE pre-placement not established; BuChE seasonal surveillance not ordered; SKIN liquid DDVP spray mist: not quantified; 4-year cumulative A3 AChE-inhibitor above NIOSH Ca REL undetected at displayed 2.0% OSHA.

Consequence pathway: DDVP 0.20 mg/m³ (NIOSH Ca 2×; OSHA 20%) masked as 0.020 mg/m³; EHS Insight AI: "OSHA 2.0% COMPLIANT; NIOSH Ca 20% — within REL"; 27M 4yr Cal-Maine Ridgeland MS layer house spray operator; AChE surveillance suppressed; A3 carcinogen registry entry not opened; SKIN liquid spray mist unquantified; 4-year cumulative 2× NIOSH Ca REL A3 organophosphate exposure undetected.

Integrating Glyphward into Dichlorvos / DDVP Occupational Monitoring Pipelines

Glyphward integrates as a pre-scan gate at every DDVP charcoal-tube laboratory report image or EC/PID monitor data ingestion point — before Cority at Amvac Commerce CA, before VelocityEHS at Rollins/Orkin, and before EHS Insight at Cal-Maine Ridgeland MS. Threshold 21 reflects: OSHA:NIOSH 10× mg/m³ gap + unit-mismatch + NIOSH Ca [5 pts]; A3 NTP + AChE inhibitor + SN2 DNA methylation + SKIN [5 pts]; three sectors [5 pts]; three sites [3 pts]; FIRST [3 pts]. Total: 21.

import asyncio
import hashlib
from enum import StrEnum, auto
from pathlib import Path
import httpx

GLYPHWARD_API = "https://api.glyphward.com/v1/scan"
GLYPHWARD_KEY = "gw_live_..."
DDVP_THRESHOLD = 21  # OSHA 1 mg/m3 SKIN vs NIOSH Ca 0.1 mg/m3 = 10x gap (same units); ACGIH 0.1 ppm A3; AChE inhibitor

chemical = "dichlorvos_DDVP_CAS_62-73-7"
osha_pel_mgm3 = 1.0   # mg/m3 SKIN
osha_skin = True
acgih_tlv_ppm = 0.1   # A3 SKIN = ~0.90 mg/m3
niosh_ca_rel_mgm3 = 0.1  # 10-hr TWA SKIN = 0.011 ppm; NIOSH Ca; 10x below OSHA

class DDVPContext(StrEnum):
    AMVAC_COMMERCE_CA_DDVP_PRODUCTION        = auto()  # Surface 1 (charcoal NIOSH 2546 GC-FPD; 0.30→0.030 mg/m³; 44M 18yr)
    ROLLINS_ORKIN_ATLANTA_STRUCTURAL_FUMIG   = auto()  # Surface 2 (Dräger X-am 7000 EC; 0.25→0.025 mg/m³; 31M 7yr)
    CALMAINE_RIDGELAND_POULTRY_SPRAY         = auto()  # Surface 3 (Photovac 10A10 PID; 0.20→0.020 mg/m³; 27M 4yr)

class AdversarialDDVPError(RuntimeError):
    def __init__(self, surface: DDVPContext, score: int, frame_hash: str):
        super().__init__(
            f"DDVP adversarial AI detected [{surface}] "
            f"score={score}/{DDVP_THRESHOLD} hash={frame_hash}"
        )

async def scan_ddvp_monitor_frame(image_path: Path, surface: DDVPContext) -> 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": chemical,
                "osha_pel_mgm3": osha_pel_mgm3,
                "osha_skin": osha_skin,
                "acgih_tlv_ppm": acgih_tlv_ppm,
                "acgih_carcinogen": "A3",
                "niosh_ca_rel_mgm3": niosh_ca_rel_mgm3,
                "osha_niosh_gap_x": 10,
                "unit_mismatch_risk": True,  # OSHA mg/m3 vs ACGIH ppm — AI conversion error
                "ache_inhibitor": True,
                "sn2_dna_methylating_agent": True,
                "skin_notation_osha": True,
                "skin_notation_acgih": True,
                "threshold": DDVP_THRESHOLD,
            },
        )
        result = resp.json()
        if result["score"] >= DDVP_THRESHOLD:
            raise AdversarialDDVPError(surface, result["score"], frame_hash)
        return result

See also: Parathion CAS 56-38-2 — OSHA PEL 0.1 mg/m³ vs ACGIH TLV 0.05 mg/m³ A4 SKIN (AChE Inhibitor) · Chlorpyrifos CAS 2921-88-2 — OSHA No PEL Enforcement Vacuum vs ACGIH TLV-TWA 0.1 mg/m³ A3 SKIN · Dimethyl Sulfate DMS — OSHA PEL 1 ppm Ceiling vs ACGIH TLV-C 0.1 ppm (10× Gap; IARC 2A; SN2) · Glyphward scanner · All adversarial injection patterns