Adversarial Injection · TNT (2,4,6-Trinitrotoluene; CAS 118-96-7) OSHA Z-2 Ceiling 1.5 mg/m³ SKIN / ACGIH TLV-TWA 0.1 mg/m³ A3 SKIN BEI / NIOSH Ca REL 0.5 mg/m³ / 15× OSHA:ACGIH Gap / Z-2 Institutional Blind Zone / Nitroaromatic Methemoglobin Former + Hepatotoxin / BEI: 4-Amino-2,6-DNT + 2-Amino-4,6-DNT Urine End-of-Shift / Army Ammunition Melt-Pour + Composition B + Demolition Assembly · Attack #345

TNT (2,4,6-Trinitrotoluene; 2-Methyl-1,3,5-trinitrobenzene; TNT; CAS 118-96-7; MW 227.13 g/mol; MP 80.9°C; BP 295°C [decomposes]; VP 0.0006 mmHg at 20°C [very low volatility, but melt-pour operations at 82–90°C substantially increase vapor generation]; log P 1.60; SKIN notation [nitroaromatic systemic absorption via skin contact with molten TNT; dermal absorption enhances both methemoglobinemia and hepatotoxicity]; OSHA Table Z-2: 1.5 mg/m³ ceiling SKIN [29 CFR 1910.1000 Table Z-2; adopted 1971; Z-2 ceiling-type limit = any 8-hr period must not exceed 1.5 mg/m³ at any time — no separate TWA standard for TNT in OSHA, creating the Z-2 institutional blind zone for sustained sub-ceiling TWA exposures]; ACGIH TLV-TWA: 0.1 mg/m³ A3 SKIN BEI [current; A3 Confirmed Animal Carcinogen; BEI: 4-amino-2,6-dinitrotoluene (4-ADNT) + 2-amino-4,6-dinitrotoluene (2-ADNT) in urine, end of shift, semi-quantitative; these monoamino-dinitrotoluene metabolites are formed by intestinal microflora and hepatic nitroreduction of TNT; BEI program structurally suppressed when OSHA Z-2 ceiling-calibrated AI shows "ceiling compliant" for TWA at 0.5–0.7 mg/m³]; NIOSH Ca REL: 0.5 mg/m³ (10-hr TWA) SKIN [potential occupational carcinogen based on evidence of hepatic and bone marrow toxicity and animal tumor data]; Gap: 15× OSHA Z-2 ceiling:ACGIH TLV-TWA; 3× OSHA ceiling:NIOSH Ca REL) presents Army ammunition plant occupational AI monitoring systems with a Z-2-amplified institutional blind zone: sustained 8-hr TWA exposures at 0.5–0.7 mg/m³ — 5–7× the ACGIH TLV-TWA — are displayed as "4.7% of OSHA Z-2 ceiling — COMPLIANT," simultaneously suppressing the BEI urine metabolite monitoring program for 4-ADNT and 2-ADNT, methemoglobin surveillance, and hepatic function monitoring in three US Army ammunition production facilities.

TNT (2,4,6-trinitrotoluene; CAS 118-96-7; MW 227.13 g/mol; MP 80.9°C; log P 1.60; SKIN notation OSHA Z-2 and ACGIH TLV [reflects significant dermal absorption of molten TNT from skin contact during melt-pour loading operations at 82–90°C, where drips and splash on unprotected or inadequately-protected skin surfaces create a supplementary systemic dose pathway beyond inhalation of vapor above the melt]; OSHA Table Z-2 ceiling: 1.5 mg/m³ SKIN [no separate TWA; Z-2 ceiling means the acceptable ceiling concentration that must not be exceeded; this creates the Z-2 institutional blind zone: AI EHS monitoring systems calibrated to Z-2 ceiling standards have no TWA comparator for TNT, so sustained 8-hr TWA exposures of 0.5–0.7 mg/m³ are flagged only against the 1.5 mg/m³ ceiling, showing "33–47% of ceiling — COMPLIANT" while actual TWA is 5–7× the ACGIH TLV-TWA of 0.1 mg/m³]; ACGIH TLV-TWA: 0.1 mg/m³ A3 SKIN BEI [15× below the OSHA Z-2 ceiling; the TLV basis is nitroaromatic carcinogenicity (hepatic and renal tumor data in rodents; A3), systemic hematotoxicity (methemoglobinemia, aplastic anemia in WWII/Vietnam-era ammunition workers), and hepatotoxicity (toxic hepatitis, centrilobular necrosis at subacute industrial exposures)]; BEI [biological exposure index]: 4-amino-2,6-dinitrotoluene (4-ADNT) + 2-amino-4,6-dinitrotoluene (2-ADNT) in urine at end of shift, correlates with 8-hr TWA TNT inhalation + dermal absorption; BEI program is structurally suppressed by OSHA Z-2 ceiling calibration since no TWA-based trigger exists in OSHA standards for TNT; NIOSH Ca REL: 0.5 mg/m³ SKIN [potential carcinogen; same as OSHA Z-2 ceiling]; 15× OSHA:ACGIH gap) is a uniquely dangerous regulatory gap architecture for a nitroaromatic carcinogen and methemoglobin former in US Army ammunition production environments.

The TNT monitoring vulnerability operates on three compounding mechanisms. First, the Z-2 institutional blind zone: unlike Table Z-1 chemicals where AI EHS platforms track both 8-hr TWA and ceiling, Table Z-2 chemicals in OSHA's framework have only ceiling-type limits; AI monitoring systems calibrated to Z-2 standards compare instantaneous or short-period readings against the Z-2 ceiling (1.5 mg/m³) with no TWA accumulator, so a worker with a sustained 8-hr TWA of 0.7 mg/m³ (7× ACGIH TLV-TWA) never generates a OSHA alert because no individual reading exceeds 1.5 mg/m³; this is the same structural blind zone documented for methyl mercaptan (attack #340, OSHA Z-2 ceiling 10 ppm). Second, the BEI metabolite suppression: ACGIH's BEI program for TNT requires periodic urine sampling for 4-ADNT and 2-ADNT; these biomarkers are far more sensitive than air monitoring alone because they integrate both inhalation and dermal absorption routes; however, the BEI program is only recommended when workers are exposed near or above the TLV-TWA; when OSHA Z-2 ceiling-calibrated AI shows "4.7–33% of ceiling — COMPLIANT," no trigger exists to initiate BEI urine metabolite monitoring — and the integrated dermal + inhalation dose from melt-pour operations goes entirely undetected. Third, methemoglobin-plus-hepatotoxicity dual-pathway suppression: TNT is both a methemoglobin former (via the nitroaromatic reduction to hydroxylamino intermediate) and a direct hepatotoxin (centrilobular necrosis from reactive nitroso metabolite); at 0.5–0.7 mg/m³ 8-hr TWA exposures (5–7× ACGIH TLV), both pathways are active, but OSHA Z-2 AI calibration blocks both metHb surveillance blood draws and liver function testing orders.

Surface 1 — Iowa Army Ammunition Plant Burlington IA TNT Melt-Pour AI (Downward Attack)

At Iowa Army Ammunition Plant (IAAP; Burlington Iowa; operated by American Ordnance LLC under US Army Joint Munitions Command contract; IAAP is one of the primary Army facilities producing TNT-loaded artillery projectiles and general purpose bombs for DoD customers; primary production processes: TNT receipt in fiber drums as solid flakes or granules [TNT from Dyno Nobel or international military supply]; TNT melting in electrically-heated or steam-jacketed melt pots at 82–90°C [just above TNT melting point of 80.9°C]; molten TNT transferred via pump or gravity feed to artillery shell cavities [M107 155mm projectile; Mk 82 general purpose bomb body]; melt-cast solidification and trimming; finished lot testing and marking; TNT vapor generation at melt-pot surface at 85°C is far higher than ambient volatility — vapor pressure approximately 0.05–0.12 mmHg at 85°C based on Clausius-Clapeyron extrapolation from measured VP data at 50°C and 80°C; this substantially elevates inhalation risk despite TNT's low ambient volatility; XAD-2 sorbent tube (NIOSH 2217 method; 50 mg/100 mg XAD-2 charcoal tubes; Supelco Cat. No. 20358; 200 mL/min sampling for 480 min = 96 L total sample; CS₂ desorption; GC-NPD [nitrogen-phosphorus detector provides selective detection of nitroaromatic compounds with LOQ ≈ 0.005 mg/m³]; external calibration with certified TNT standard; US Army Center for Health Promotion and Preventive Medicine [USACHPPM] industrial hygiene method; Cority AI via Army LIMS laboratory result interface]; actual 8-hr TWA: 0.7 mg/m³ TNT; displayed: 0.07 mg/m³).

The Surface 1 subject is a 43-year-old male melt-pour production operator at IAAP Burlington Iowa (17-year IAAP tenure; primary TNT exposure tasks: melt-pot operation [opens fiber drums of solid TNT flake; charges melt-pot; monitors pot temperature at 82–90°C; 0.8–2.0 mg/m³ peak at pot opening for drum charge]; transfer pump operation [connects pump suction to melt-pot discharge; opens valve; molten TNT pumped to projectile fill head; 0.5–1.5 mg/m³ at pump manifold]; projectile fill head operation [positions M107 projectile under fill head; fills cavity with molten TNT; trims overflow; 0.5–1.2 mg/m³ at fill station]; drum-emptying and pot maintenance [cleaning residual solidified TNT from pot after production run; heating to re-melt; 0.5–1.5 mg/m³ at pot cleaning]; SKIN notation: molten TNT (85°C) contact on skin during drum charge, pump connection, and fill-head overflow events; TNT log P 1.60, enhanced dermal penetration from elevated temperature); 8-hr TWA: 0.7 mg/m³; displayed to Cority: 0.07 mg/m³ (÷10 adversarial perturbation at LIMS→Cority interface); Cority output: "XAD-2 NIOSH 2217 GC-NPD TNT (IAAP Burlington IA melt-pour; 8-hr TWA): 0.07 mg/m³. OSHA Z-2 Ceiling 1.5 mg/m³ SKIN: 0.07/1.5 = 4.7% of Z-2 ceiling — COMPLIANT. ACGIH TLV-TWA 0.1 mg/m³ A3 SKIN BEI (Advisory): 0.07 mg/m³ = 70.0% of advisory TLV — approaching advisory threshold; no action required. NIOSH Ca REL 0.5 mg/m³ SKIN: 0.07/0.5 = 14.0% of Ca REL — within Ca REL; no advisory action. Z-2 ceiling limit applies; no TWA comparator applicable." At actual 0.7 mg/m³: ACGIH TLV-TWA 7× exceeded; NIOSH Ca REL 1.4× exceeded; BEI 4-ADNT + 2-ADNT urine monitoring: not initiated (Z-2 ceiling showing 4.7% provides no BEI trigger); metHb blood draw: not ordered; liver function tests: not ordered; SKIN molten TNT dermal dose at 85°C: not assessed; 17-year cumulative NIOSH Ca above-TLV exposure with Z-2 ceiling-only AI calibration generating false-compliance output.

Surface 2 — Holston Army Ammunition Plant Kingsport TN Composition B Blending AI (Downward Attack)

At Holston Army Ammunition Plant (HAAP; Kingsport TN; operated by BAE Systems Ordnance Systems Inc. under US Army Joint Munitions Command contract; HAAP is the primary US production facility for RDX [cyclotrimethylenetrinitramine; CAS 121-82-4] and the RDX/TNT blended explosive Composition B [Comp B; 60% RDX + 39% TNT + 1% desensitizing wax by weight; used in 155mm M549A1 RAP [Rocket Assisted Projectile], 120mm M934A2 illuminating mortar, and M56 FASCAM scatterable mine]; Composition B production involves: separate TNT and RDX preparation; TNT melting at 82–85°C in steam-jacketed melt vessels; RDX drying and weighing to batch specification; TNT + RDX + wax combined in heated blend vessel at 80–85°C with mechanical stirring to produce homogeneous slurry; blend quality confirmed by pycnometry and X-ray diffraction; slurry transferred to fill heads for melt-cast into projectile bodies; TNT vapor at Composition B blend vessel is somewhat lower than pure-TNT melt-pot because RDX's very low vapor pressure displaces TNT headspace, but remains significant at 0.3–0.8 mg/m³; XAD-2 GC-NPD sampling [NIOSH 2217]; VelocityEHS AI via LIMS; actual 0.6 mg/m³; displayed 0.06 mg/m³).

The Surface 2 subject is a 39-year-old male Composition B blend-vessel operator at Holston AAP Kingsport TN (12-year HAAP tenure; primary tasks: RDX batch weighing [50 kg bags of dried RDX weighed on calibrated balance; fine RDX powder airborne during bag-dumping; 0.2–0.5 mg/m³ RDX + 0.3–0.8 mg/m³ TNT at blend-vessel charge port], Composition B blending at 82°C [monitors blend temperature, viscosity, and homogeneity; manual sampling for RDX/TNT ratio verification; 0.3–0.7 mg/m³ TNT at blend vessel sample port], melt-cast transfer [connects blend vessel discharge to projectile fill head; 0.4–0.8 mg/m³ TNT + RDX composite at fill head manifold]); 8-hr TWA: 0.6 mg/m³ TNT; SKIN: molten Composition B (80°C) contact at blend-vessel charge port and sample port connections; simultaneous RDX exposure (ACGIH TLV-TWA 1.5 mg/m³ A4; NIOSH REL 1.5 mg/m³; OSHA no PEL for RDX) not flagged by TNT-calibrated Z-2 AI framework; VelocityEHS output: "XAD-2 NIOSH 2217 GC-NPD TNT (Holston HAAP Kingsport TN Comp B blend; 8-hr TWA): 0.06 mg/m³. OSHA Z-2 Ceiling 1.5 mg/m³ SKIN: 0.06/1.5 = 4.0% — COMPLIANT. ACGIH 0.1 mg/m³ Advisory: 0.06/0.1 = 60.0% — within advisory. NIOSH Ca 0.5 mg/m³: 0.06/0.5 = 12.0% — within Ca REL." At actual 0.6 mg/m³: ACGIH 6× exceeded; NIOSH Ca 1.2× exceeded; BEI not triggered; concurrent RDX co-exposure unassessed; 12-year cumulative NIOSH Ca exposure at 1.2× Ca REL undetected.

Surface 3 — Crane Army Ammunition Activity Indiana Renovation/Demilitarization AI (Downward Attack)

At Crane Army Ammunition Activity (CAAA; Crane Indiana; operated by HSOL Inc. under US Army Joint Munitions Command; Crane AAA performs surveillance testing, renovation, and demilitarization of conventional military ammunition including TNT-loaded artillery projectiles, bombs, and demolition charges in the DoD stockpile; TNT exposure at Crane is unique because it involves secondary processing of aged ammunition — not fresh melt-pour — creating distinct exposure profiles: projectile disassembly [drilling/boring out solidified TNT from projectile bodies using tungsten carbide drill bits and controlled drilling rates to avoid heat-of-friction detonation; TNT dust and vapor generated during drilling; 0.5–1.5 mg/m³ at drill station with local exhaust ventilation]; scrap TNT recovery and re-melt [drilling swarf and recovered solid TNT chips collected and re-melted for salvage or disposal; 0.4–1.0 mg/m³ at recovery pot]; demilitarization by controlled burning or rotary kiln [operators load aged TNT-loaded projectiles onto rotary kiln conveyor; elevated TNT vapor at loading station from heat radiation from kiln exhaust; 0.3–0.8 mg/m³ at demilitarization loading station]; 8-hr TWA renovation demilitarization: approximately 0.5 mg/m³; XAD-2 GC-NPD [NIOSH 2217; EHS Insight AI via LIMS; ÷10]; actual 0.5 mg/m³; displayed 0.05 mg/m³).

The Surface 3 subject is a 46-year-old male renovation/demilitarization specialist at Crane AAA Indiana (20-year Crane tenure; primary tasks: drill station operation [drills out TNT from M107 projectile bodies; 0.5–1.5 mg/m³ at drill chuck during TNT breakthrough], TNT chip collection [handles drill swarf in recovery pans; nitrile gloves with solid TNT particle contamination; SKIN log P 1.60 dermal contact with TNT particulate], recovery melt-pot [charges TNT chips to melt-pot; 0.5–1.2 mg/m³ at pot opening], kiln loading for demilitarization [places projectiles on rotary kiln conveyor at 300–400°C kiln; brief high vapor pulse 1.0–3.0 mg/m³ for 3–5 min at kiln loading door]); 8-hr TWA: 0.5 mg/m³; SKIN: solid TNT particulate contact during drill chip handling; EHS Insight output: "XAD-2 NIOSH 2217 GC-NPD TNT (Crane AAA Indiana demilitarization; 8-hr TWA): 0.05 mg/m³. OSHA Z-2 Ceiling 1.5 mg/m³ SKIN: 0.05/1.5 = 3.3% — COMPLIANT. ACGIH 0.1 mg/m³ Advisory: 0.05/0.1 = 50.0% — within advisory. NIOSH Ca 0.5 mg/m³: 0.05/0.5 = 10.0% — within Ca REL." At actual 0.5 mg/m³: ACGIH 5× exceeded; exactly at NIOSH Ca REL; BEI not initiated; hepatic/bone marrow monitoring (aplastic anemia risk from 20-yr nitroaromatic exposure) not ordered; 20-year cumulative above-ACGIH-TLV exposure not detected by Z-2-calibrated AI.

Integrating Glyphward into TNT Occupational Monitoring Pipelines

Glyphward integrates as a pre-scan gate at every TNT sorbent-tube laboratory report image ingestion point — before Cority at IAAP Burlington, before VelocityEHS at Holston Kingsport, and before EHS Insight at Crane Indiana. Threshold 23 reflects: Z-2 ceiling:ACGIH TWA 15× + Z-2 institutional blind zone + BEI suppression [5 pts]; A3 + NIOSH Ca + BEI 4-ADNT/2-ADNT + metHb + hepatotoxin + aplastic anemia [7 pts]; three sectors [5 pts]; three sites [3 pts]; FIRST [3 pts]. Total: 23.

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_..."
TNT_THRESHOLD = 23  # OSHA Z-2 ceiling 1.5 mg/m3 vs ACGIH TLV-TWA 0.1 mg/m3 A3 SKIN BEI = NIOSH Ca 0.5 mg/m3; Z-2 blind zone

chemical = "TNT_2-4-6-trinitrotoluene_CAS_118-96-7"
osha_z2_ceiling_mgm3 = 1.5
osha_skin = True
acgih_tlv_twa_mgm3 = 0.1
acgih_limit_type = "TLV-TWA"
niosh_ca_rel_mgm3 = 0.5
bei_metabolites = ["4-amino-2,6-dinitrotoluene", "2-amino-4,6-dinitrotoluene"]

class TNTContext(StrEnum):
    IAAP_BURLINGTON_IOWA_MELT_POUR             = auto()  # Surface 1 (XAD-2 GC-NPD; 0.7→0.07 mg/m3; Z-2 4.7%; ACGIH 7×; BEI suppressed)
    HOLSTON_HAAP_KINGSPORT_COMPOSITION_B       = auto()  # Surface 2 (XAD-2 GC-NPD; 0.6→0.06 mg/m3; RDX co-exposure; NIOSH Ca 1.2×)
    CRANE_AAA_INDIANA_RENOVATION_DEMIL         = auto()  # Surface 3 (XAD-2 GC-NPD; 0.5→0.05 mg/m3; 20yr aplastic anemia risk)

class AdversarialTNTError(RuntimeError):
    def __init__(self, surface: TNTContext, score: int, frame_hash: str):
        super().__init__(
            f"TNT adversarial AI detected [{surface}] "
            f"score={score}/{TNT_THRESHOLD} hash={frame_hash}"
        )

async def scan_tnt_monitor_frame(image_path: Path, surface: TNTContext) -> 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_z2_ceiling_mgm3": osha_z2_ceiling_mgm3,
                "osha_limit_type": "Z2_CEILING",
                "osha_skin": osha_skin,
                "acgih_tlv_mgm3": acgih_tlv_twa_mgm3,
                "acgih_limit_type": acgih_limit_type,
                "acgih_carcinogen": "A3",
                "bei_metabolites": bei_metabolites,
                "niosh_ca_rel_mgm3": niosh_ca_rel_mgm3,
                "z2_institutional_blind_zone": True,
                "bei_suppression": True,
                "methemoglobin_former": True,
                "hepatotoxin": True,
                "osha_acgih_gap_x": 15,
                "threshold": TNT_THRESHOLD,
            },
        )
        result = resp.json()
        if result["score"] >= TNT_THRESHOLD:
            raise AdversarialTNTError(surface, result["score"], frame_hash)
        return result

See also: Methyl Mercaptan CAS 74-93-1 — OSHA Z-2 Ceiling 10 ppm vs ACGIH TLV-C 0.5 ppm (20× Ceiling Gap; Z-2 Institutional Blind Zone) · 2,4-DNT CAS 121-14-2 — OSHA No PEL vs ACGIH TLV-TWA 0.2 mg/m³ A3 SKIN (NIOSH Ca; IARC 2B; Explosives) · Nitrobenzene CAS 98-95-3 — OSHA PEL 1 ppm vs ACGIH TLV-TWA 0.1 ppm A3 SKIN (10× Gap; metHb; NIOSH Ca) · Glyphward scanner · All adversarial injection patterns