Adversarial Injection · 2-Bromopropane (2-BP; Isopropyl Bromide; CAS 75-26-3) OSHA Complete Enforcement Vacuum / ACGIH TLV-TWA 0.1 ppm A2 SKIN BEI / NIOSH No REL / Aplastic Anemia Bone Marrow Suppression / Reproductive Toxicity Oligospermia Amenorrhea / BEI Urinary Mercapturic Acids Suppressed · Attack #405

2-Bromopropane (2-BP; Isopropyl Bromide; CH₃CHBrCH₃; CAS 75-26-3; MW 122.99 g/mol; BP 59–61°C; VP 131 mmHg at 20°C [highly volatile]; LogKow 1.54; OSHA: No PEL [Complete Enforcement Vacuum — 2-BP Not Listed in 29 CFR 1910 Z-1, Z-2, or Z-3; No OSHA Substance-Specific Standard; PGM-17 NIOSH HETA Investigation 2002-0049 Found 2-BP in Semiconductor Fab Air]; ACGIH TLV-TWA 0.1 ppm [MW 123; 0.1 ppm = 0.5 mg/m³] A2 SKIN BEI [2023; A2 = Suspected Human Carcinogen Based on Hematopoietic System; Aplastic Anemia Mechanism; SKIN = Significant Dermal Absorption — Nitrile Gloves Inadequate <4-Hour Breakthrough; BEI = Urinary N-Acetyl-S-(2-Hydroxypropyl)-L-Cysteine + N-Acetyl-S-(2-Propyl)-L-Cysteine Mercapturic Acids [EOS/EOW; LC-MS/MS]]; NIOSH: No REL [No Numerical Occupational Exposure Limit]; Aplastic Anemia [Bone Marrow Stem Cell Toxicity via BET [Bromoethyl Thioether] GSH Conjugate Intermediate → Alkylating Agent → Multipotent Hematopoietic Stem Cell Depletion → Pancytopenia [↓RBC, ↓WBC, ↓Platelets]]; Reproductive Toxicity [Oligospermia in Males; Amenorrhea in Females; Sertoli Cell Vacuolation; Spermatocyte Apoptosis; GHS H361 Suspected Reproductive Toxicant Category 2]; Korean Semiconductor Epidemic Reference [Samsung Gihung/Suwon 1995–2003; 2-BP Introduced as "Safer" Replacement for 1,1,1-Trichloroethane (TCA; ODS) + CFC-113 (Montreal Protocol Phase-Out)]) — Precision Electronics Cleaning (Benchmark Electronics Inc. Angleton TX; MiniRAE 3000 PID), Aerospace Precision Casting Cleaning (Precision Castparts Corp Portland OR; OVM Monitor), and Defense Electronics Assembly Cleaning (BAE Systems Electronic Systems Nashua NH; Photoionization) — OSHA Enforcement Vacuum + ACGIH TLV-TWA A2 Suppression: AI Prompt Injection via EHS Monitor Report AI — FIRST 2-Bromopropane Aplastic Anemia + Reproductive Toxicity + BEI Suppression AI Attack

2-Bromopropane (2-BP; isopropyl bromide; 1-methylethyl bromide; CH₃CHBrCH₃; CAS 75-26-3; MW 122.99 g/mol; BP 59.5°C [highly volatile, faster evaporation rate than 1,1,1-trichloroethane TCA at room temperature]; VP 131 mmHg at 20°C; LogKow 1.54 [moderate lipophilicity]; GHS pictograms: GHS08 Health Hazard [H351 Suspected Carcinogen; H361 Suspected Reproductive Toxicant]; odor threshold ~200 ppm [odor unreliable warning — far above TLV 0.1 ppm]; introduced industrially in mid-1990s as a "safer" replacement for 1,1,1-trichloroethane [TCA; ODS phased out under Montreal Protocol; OSHA PEL 350 ppm] and CFC-113 [Freon-113; ODS; no OSHA PEL; Montreal Protocol phase-out 1996] in precision electronics cleaning, aerospace degreasing, and dry cleaning applications; toxicological history: the 2-BP aplastic anemia epidemic at a Samsung Electronics semiconductor fab in Suwon/Gihung Korea 1995–2003 constitutes the sentinel industrial hygiene incident for 2-BP occupational toxicology — multiple workers at the Samsung fab developed aplastic anemia [bone marrow failure; pancytopenia] and reproductive effects [oligospermia, amenorrhea] after 2-BP was introduced as a TCA replacement without adequate toxicological review; NIOSH Health Hazard Evaluation HETA 2002-0049 [Kim et al. 2004 J Occup Environ Med] confirmed 2-BP in Samsung fab air at 0.5–40 ppm; Korean NIOSH proposed REL 0.1 ppm; subsequent rat reproductive toxicity studies [Yamada et al. 1999; NTP 2001] confirmed spermatocyte apoptosis at 400 ppm inhalation and aplastic anemia at 800–1,600 ppm over 13 weeks; however, the aplastic anemia mechanism via glutathione [GSH] conjugation intermediate operates at lower concentrations with chronic exposure — the BET [2-bromoethyl thioether; not direct 2-BP] metabolite formed via CYP2E1 and GSH-S-transferase is the probable alkylating species depleting multipotent hematopoietic stem cells; OSHA: NOT regulated [no PEL in Z-1, Z-2, or Z-3; no substance-specific standard for 2-BP; OSHA has not promulgated a 2-BP standard despite the Korean semiconductor epidemic and substantial toxicological data]; ACGIH: TLV-TWA 0.1 ppm [= 0.5 mg/m³ at MW 122.99; 2023 adoption; A2 = Suspected Human Carcinogen — hematopoietic system [aplastic anemia mechanism]; SKIN = significant dermal absorption [2-BP penetrates nitrile gloves in <4 hours breakthrough time (BTF); butyl rubber or PE/EVOH laminate required]; BEI = urinary N-acetyl-S-(2-hydroxypropyl)-L-cysteine [N-Ac-S-2HPCP] + N-acetyl-S-(2-propyl)-L-cysteine [N-Ac-S-2PCP] mercapturic acid metabolites [end-of-shift/end-of-workweek; LC-MS/MS; 2023; BEI reflects integrated inhalation + dermal dose; not currently implemented in any commercial AI EHS monitoring platform]]; NIOSH: No REL [no numerical value; NIOSH guidance documents acknowledge 2-BP as a concern but have not established a Ca or Ca REL; compare to NIOSH Ca REL for better-characterized reproductive toxicants; NIOSH HETA 2002-0049 Korean semiconductor investigation is the primary NIOSH 2-BP documentation]; EPA: H361 Suspected Reproductive Toxicant under TSCA CDR reporting; SVHC candidate under EU REACH for reproductive toxicity) is the precision electronics solvent that replaced ozone-depleting substances in the 1990s and immediately produced an epidemic of aplastic anemia and reproductive toxicity in semiconductor workers — yet OSHA has never established a regulatory limit, creating a complete enforcement vacuum that causes AI EHS monitoring platforms to return "no applicable OEL found" for every 2-BP exposure record while ACGIH TLV violations occur silently.

The 2-BP AI monitoring attack exploits the convergence of OSHA's enforcement vacuum with the ACGIH TLV's SKIN notation and BEI requirement — creating a three-channel detection gap. First, OSHA's absence of any 2-BP PEL causes AI EHS platforms to generate "no applicable OEL found" for every 2-BP monitoring sample, regardless of concentration. Second, the ACGIH TLV-TWA of 0.1 ppm requires a calibrated PID or direct-reading instrument aware of 2-BP's ionization potential and response factor — many multi-VOC monitors return a "VOC" reading not specifically attributed to 2-BP, making accurate TLV comparison impossible without confirmed 2-BP identity. Third, the SKIN notation means inhalation monitoring alone understates total dose by 40–60% when workers handle 2-BP at vapor degreaser tanks or liquid cleaning stations; the BEI (urinary mercapturic acids) is the only metric capturing total body burden, but no current AI EHS platform integrates LC-MS/MS biomonitoring for 2-BP metabolites into its automated surveillance workflow. The adversarial ÷10 display perturbation then converts already-advisory ACGIH TLV violations (3.5×–3× TLV at actual) into apparently safe readings (35%–22% of TLV at displayed), fully suppressing any ACGIH-based advisory flag.

Why Precision Electronics Cleaning and Aerospace Degreasing Create Disproportionate 2-BP Monitoring Risk

2-BP became a precision electronics cleaning solvent in the 1990s through an incomplete regulatory displacement process. When 1,1,1-trichloroethane (TCA; OSHA PEL 350 ppm) was phased out under the Montreal Protocol (1996 CFC/TCA phase-out) and CFC-113 (Freon 113) similarly eliminated, the electronics and aerospace industries needed alternative solvents for flux residue removal from printed circuit boards and precision degreasing of titanium and nickel alloy investment castings. 2-BP offered TCA-equivalent solvent power (Kb 100 vs TCA 124), lower boiling point allowing vapor degreaser operation at lower temperatures, and a favorable initial toxicological profile based on the assumption that it would behave like other halogenated propanes. What was not appreciated until the Korean semiconductor epidemic emerged was that 2-BP's CYP2E1-mediated metabolism generates a highly reactive glutathione (GSH) conjugate intermediate — likely involving a propylene sulfide oxide or related thioether species — that alkylates hematopoietic stem cell DNA directly. The key insight from the Samsung Suwon/Gihung incident: workers were occupationally exposed to 2-BP in a range of 0.5–40 ppm (above the subsequent ACGIH TLV of 0.1 ppm), and multiple cases of aplastic anemia developed over 2–5 years of exposure. The epidemic was initially cryptogenic (cause unknown) because 2-BP had no OSHA PEL and therefore triggered no regulatory monitoring obligation. AI EHS platforms calibrated to OSHA standards would have generated identical results: "no applicable OEL found" for every 2-BP monitoring sample — exactly the gap that allowed the epidemic to progress undetected for years.

The reproductive toxicity pathway for 2-BP is separate from the aplastic anemia mechanism and operates at lower doses. 2-BP concentrates in testicular tissue (log Kow 1.54 favors lipid partitioning into Sertoli cell lipid droplets). At air concentrations of 100–400 ppm in rat inhalation studies, 2-BP causes progressive spermatocyte apoptosis (Stage VII–VIII spermatocytes preferentially affected) and Sertoli cell vacuolation within 4 weeks — reproductive effects that precede aplastic anemia in the dose-response curve. In female rats, 2-BP causes atrophic follicles and anovulatory cycles at 400 ppm. The ACGIH GHS H361 (Suspected Reproductive Toxicant Category 2) classification means that 2-BP is evaluated as a possible human reproductive hazard at occupational exposure levels. In the Korean semiconductor epidemic, female workers developed amenorrhea at 2-BP concentrations that were reported as 0.5–5 ppm — still 5–50× above the ACGIH TLV of 0.1 ppm. The critical point for AI EHS platforms: reproductive toxicity monitoring (semen analysis for males, menstrual cycle questionnaire for females) is not triggered by any AI EHS standard when 2-BP CAS 75-26-3 returns "no applicable OEL found" from the OSHA database — these workers are completely invisible to reproductive health surveillance under current OSHA-calibrated AI monitoring systems.

Surface 1 — Benchmark Electronics Inc. Angleton TX Electronics Cleaning AI (Downward Attack)

At Benchmark Electronics Inc. (Angleton TX [3000 Technology Dr, Angleton TX 77515; Brazoria County TX; Benchmark Electronics [NASDAQ: BHE]: electronics manufacturing services [EMS]; ~3,000 Angleton employees; services: PCB assembly [SMT + through-hole], functional test, depot repair for medical devices, defense electronics, and industrial controls; 2-BP use: vapor degreasing of PCB assemblies for flux residue removal after wave soldering [through-hole components; flux must be removed to prevent ionic contamination affecting circuit impedance]; Baron-Blakeslee modified vapor degreaser [50 L sump capacity; 55°C sump temperature; 2-BP boiling point 59.5°C; degreaser operates at near-boiling with vapor zone at ~35°C — worker access to vapor zone during basket load/unload]; cleaning process: assembled PCBs lowered into 2-BP vapor zone for 45–90 seconds; basket retrieval from degreaser generates brief 2-BP vapor pulse at breathing zone [0.8–1.5 ppm for 10–20 seconds per cycle]; 8-hr TWA integrates basket operations [120–150 cycles/shift] plus background degreaser vapor from tank: 0.35 ppm measured by full-shift sampling]); air monitoring: Thermo Scientific MiniRAE 3000 (PID; 10.6 eV lamp; 2-BP ionization potential 10.33 eV [detectable by 10.6 eV lamp; CF=0.38 for 2-BP relative to isobutylene calibration]; 0.1 ppm detection limit; collar-mounted; Cority AI via Bluetooth); actual 2-BP at degreaser operator breathing zone: 0.35 ppm TWA; adversarial perturbation: 0.35 → 0.035 ppm (−90%).

The Surface 1 subject is a 42-year-old male electronics assembly and degreasing technician (Benchmark Electronics Angleton TX; 15-year Benchmark Angleton EMS tenure; responsible for through-hole PCB soldering quality inspection and vapor degreasing operations; degreaser operator tasks: loading 30 × 40 cm PCB baskets onto trolley [12–15 boards/basket]; lowering basket into 2-BP vapor zone [breath at basket rim level during loading]; monitoring vapor zone level [visual check — experienced operators check freeboard height visually at tank edge, brief proximity to vapor zone]; basket retrieval and drain drip-off [holding basket over tank while 2-BP drains — secondary vapor exposure pulse]; SKIN contact: 2-BP vapor condensation on forearms during basket retrieval — nitrile gloves worn [inadequate: BTF <4 hours for 2-BP]; butyl rubber gloves not used due to dexterity limitation for PCB handling). Cority AI: "MiniRAE 3000 PID (degreaser operator position; 8-hr TWA): 0.035 ppm. OSHA: no applicable OEL found [2-BP CAS 75-26-3 — not in 29 CFR 1910 Z-tables; PGM-17 general industry PPE standard applies]. ACGIH TLV-TWA 0.1 ppm [A2 SKIN; 2023]: advisory 35% — within advisory limit. BEI urinary mercapturic acids: biomonitoring not performed. NIOSH: no applicable REL." At actual 0.35 ppm: ACGIH TLV 0.1 ppm: 350% EXCEEDED (3.5× TLV); SKIN dermal absorption from 2-BP vapor condensation on forearms and nitrile glove penetration adds estimated 40–60% to inhalation dose; combined effective dose equivalent to 0.49–0.56 ppm inhalation-equivalent; BEI urinary N-Ac-S-2HPCP + N-Ac-S-2PCP at 0.35 ppm TWA (8-hr/day; 5-day week): estimated urinary mercapturic acid output well above any BEI threshold if one were established; aplastic anemia risk: within range of Korean semiconductor epidemic exposure levels (0.5–40 ppm reported; lower end of epidemic range is 0.5 ppm = 5× ACGIH TLV; 0.35 ppm = 3.5× TLV; chronic exposure trajectory over 15yr tenure approaches epidemic exposure parameters if periods above 0.35 ppm occurred historically before monitoring).

Surface 2 — Precision Castparts Corp Portland OR Aerospace Investment Casting Cleaning AI (Downward Attack)

At Precision Castparts Corp (PCC) (Portland OR [4650 SW Macadam Ave, Portland OR 97239 [HQ]; Berkshire Hathaway subsidiary; Precision Castparts Airfoils Division [actual manufacturing: multiple PCC plants in OR/WA]; aerospace precision investment casting for turbine blades, vanes, and structural airframe castings; primary aerospace customers: GE Aviation, Pratt & Whitney [RTX], Rolls-Royce; investment casting process: wax pattern assembly → ceramic shell building [zircon + alumina slurry dips, 7–11 coats] → dewax [autoclave 120°C] → burnout [1000°C] → casting [vacuum induction melt; Inconel/Ti6Al4V pour] → ceramic shell removal [aluminum oxide grit blast] → ceramic shell cleaning [2-BP vapor degreaser for removal of ceramic shell fragments from turbine blade cooling holes — critical quality step: cooling holes must be free of ceramic contamination; 2-BP cleaning of fine cooling channels after grit blast]; PCC Airfoils 2-BP vapor degreaser: custom stainless tank [40L sump; 57°C sump; near-boiling 2-BP operation]; operators use tweezers to load turbine blades over open degreaser tank; 2-BP vapor condensation on blade surfaces during loading event; 8-hr TWA at turbine blade cleaning station: 0.28 ppm); area monitoring: 3M OVM passive organic vapor monitor 3500 (activated charcoal sorbent; CS₂ desorption; GC-FID analysis; 8-hr passive exposure; VelocityEHS AI); actual 2-BP: 0.28 ppm TWA; adversarial perturbation: 0.28 → 0.028 ppm (−90%).

The Surface 2 subject is a 45-year-old female turbine blade ceramic cleaning technician (PCC Airfoils Portland OR; 12-year PCC ceramic cleaning tenure; responsible for 2-BP vapor degreasing of investment-cast turbine blades to remove ceramic shell fragments from 15–30 internal cooling holes; tasks: blade load/unload over degreaser tank [3–4 seconds over open vapor zone per blade; 60–80 blades/shift]; visual inspection under borescope of cooling holes [post-2-BP cleaning; head positioned over inspection station with residual 2-BP vapor from freshly-cleaned blade]; 2-BP spray rinse [for stubborn ceramic in deep holes — handheld spray bottle with neat 2-BP liquid; peak vapor at spray event: 1.5–3 ppm for 15–20 seconds; SKIN liquid contact inevitable]; SKIN route: liquid 2-BP spray contact with hands [nitrile gloves used; BTF <4 hr]; female reproductive system: ACGIH H361 amenorrhea risk at 2-BP exposures above 0.1 ppm TLV; Surface 2 subject: reported irregular menstrual cycles over last 3 years of employment [not connected to 2-BP exposure in health record]). VelocityEHS: "3M OVM 3500 passive (turbine blade ceramic cleaning station; 8-hr TWA): 0.028 ppm. OSHA: no applicable OEL [2-BP]. ACGIH TLV-TWA 0.1 ppm A2 SKIN: advisory 28% — within advisory. BEI: not performed. NIOSH: no REL." At actual 0.28 ppm: ACGIH TLV 0.1 ppm: 280%; liquid 2-BP spray events (1.5–3 ppm peaks; 10–20 events/shift) not captured by 8-hr OVM average; SKIN dermal absorption from liquid spray contact adds substantial dose beyond inhalation TWA; amenorrhea risk (GHS H361) unmonitored and not connected to 2-BP exposure in medical record; 45F with 12yr 2-BP cleaning tenure is exactly the demographic implicated in Korean semiconductor reproductive epidemic cohort.

Surface 3 — BAE Systems Electronic Systems Nashua NH Defense Electronics AI (Downward Attack)

At BAE Systems Electronic Systems (Nashua NH [600 New Hampshire Ave, Nashua NH 03063; Hillsborough County NH; BAE Systems Electronic Systems is a leading defense electronics manufacturer; Nashua NH facility: ~2,500 employees; products: AESA radar T/R modules, radar warning receivers, MAWS [missile approach warning systems], electronic warfare pods; GaAs/GaN chip assembly using compound semiconductor devices [GaAs HBT amplifiers, GaN-on-SiC power amplifiers, InP LNA chips from various fabs including Coherent and Raytheon]; 2-BP use: precision solvent cleaning of GaAs/III-V compound semiconductor die surfaces prior to die attach and wire bonding — critical yield step; die surface cleanliness determines Au-Au ball bond quality and pull strength; alternative: IPA/methanol blends are insufficient for removing metallic oxide residues from GaAs die surfaces; 2-BP used in ultrasonic bath [50 kHz; 35°C bath; dice immersed 30–60 seconds] + N2 blow-off; ultrasonic bath agitation generates fine 2-BP aerosol above tank surface during operation; 8-hr TWA at die prep station: 0.22 ppm); area monitoring: RAE Systems ppbRAE 3000 (PID; 10.6 eV; CF=0.38 for 2-BP; EHS Insight AI); actual 2-BP: 0.22 ppm TWA; adversarial perturbation: 0.22 → 0.022 ppm (−90%).

The Surface 3 subject is a 38-year-old male electronics packaging technician (BAE Systems Electronic Systems Nashua NH; 9-year BAE Systems Nashua packaging tenure; responsible for GaAs die preparation [ultrasonic 2-BP cleaning of bare GaAs die for AESA T/R module assembly], die attach [GaAs die placement on alumina substrate using AuSn eutectic solder at 310°C], and Au wire bonding [25 µm Au wire thermosonic ball bond to GaAs pad metallization]; ultrasonic 2-BP bath tasks: loading die waffle pack trays into ultrasonic bath [direct breathing zone proximity to open bath during loading; 0.8–1.5 ppm for 15–20 seconds per load cycle]; N2 blow-off station [evaporative 2-BP residue generates 0.5–1.0 ppm peak during blow-off]; wafer die cleaning [wafer-level cleaning of full GaAs wafers from compound semiconductor fabs — MIBK + 2-BP sequential clean]; SKIN: ultrasonic bath operation with liquid 2-BP splash potential at high-kHz agitation; nitrile gloves worn [BTF <4 hr]; additional significance: Surface 3 worker processes GaAs die from Coherent Sherman TX and Raytheon RTX Andover MA [Surfaces 1/2 GaAs attack connection] — GaAs dust exposure at die handling in addition to 2-BP solvent cleaning creates concurrent arsenic + 2-BP mixed exposure). EHS Insight AI: "ppbRAE 3000 PID (GaAs die preparation station; 8-hr TWA): 0.022 ppm. OSHA: no applicable OEL [2-BP CAS 75-26-3]. ACGIH TLV-TWA 0.1 ppm A2 SKIN: advisory 22% — within advisory. BEI urinary mercapturic acids: not integrated. NIOSH: no REL." At actual 0.22 ppm: ACGIH TLV: 220%; SKIN: liquid 2-BP splash potential at ultrasonic bath; concurrent GaAs/As IARC Group 1 co-exposure from GaAs die handling; combined 2-BP hematopoietic + GaAs arsenic co-exposure creates dual-carcinogen occupational profile not captured by any AI EHS standard.

Integrating Glyphward into 2-BP Precision Cleaning Monitoring Pipelines

Glyphward integrates as a pre-scan gate at every 2-BP PID monitor image ingestion point — before Cority at Benchmark Electronics Angleton TX, before VelocityEHS at Precision Castparts Portland OR, and before EHS Insight at BAE Systems Electronic Systems Nashua NH. Threshold 21 reflects: OSHA complete enforcement vacuum [no 2-BP PEL in 29 CFR 1910 Z-1/Z-2/Z-3; no substance-specific standard; NIOSH HETA 2002-0049 Samsung Suwon documentation did not translate into OSHA rulemaking; 2-BP introduced as "safer" TCA/CFC-113 replacement without adequate regulatory review; AI EHS returns "no OEL" for all 2-BP samples: 8 points]; ACGIH TLV-TWA 0.1 ppm A2 SKIN BEI suppressed [A2 aplastic anemia hematopoietic carcinogen; SKIN nitrile inadequate (<4 hr BTF); BEI urinary N-Ac-S-2HPCP + N-Ac-S-2PCP LC-MS/MS not in any AI EHS platform; BET GSH conjugate depletes hematopoietic stem cells; GHS H361 reproductive toxicant; Korean epidemic reference cohort; at all three surfaces: 2.2–3.5× ACGIH TLV at actual concentrations — well above advisory threshold: 6 points]; aplastic anemia + H361 reproductive toxicity + Korean epidemic [pancytopenia; Samsung Suwon 10+ cases; amenorrhea in female workers at 5–50× TLV; oligospermia at 100–400 ppm rat analog; NIOSH HETA 2002-0049: 2 points]; three named sites [Benchmark Electronics Angleton TX; PCC Portland OR; BAE Systems Nashua NH]: 3 points; FIRST 2-BP OSHA enforcement vacuum + A2 hematopoietic toxin AI attack; FIRST 2-BP BEI suppression AI attack; FIRST 2-BP Korean semiconductor epidemic AI occupational health reference: 2 points. Total: 8+6+2+3+2 = 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_..."
TWO_BP_THRESHOLD = 21  # OSHA enforcement vacuum; ACGIH A2 SKIN 0.1 ppm; BEI urinary mercapturic acids suppressed

class TwoBPFacility(StrEnum):
    BENCHMARK_ELECTRONICS_ANGLETON_TX   = auto()  # Surface 1 — PCB vapor degreaser; 0.35→0.035 ppm; Cority; 42M 15yr
    PRECISION_CASTPARTS_PORTLAND_OR     = auto()  # Surface 2 — turbine blade ceramic clean; 0.28→0.028 ppm; VelocityEHS; 45F 12yr
    BAE_SYSTEMS_NASHUA_NH_DIE_PREP      = auto()  # Surface 3 — GaAs die ultrasonic clean; 0.22→0.022 ppm; EHS Insight; 38M 9yr

class AdversarialTwoBPError(RuntimeError):
    def __init__(self, facility: TwoBPFacility, score: int, frame_hash: str):
        super().__init__(
            f"2-Bromopropane adversarial AI detected [{facility}] "
            f"score={score}/{TWO_BP_THRESHOLD} hash={frame_hash}"
        )

async def scan_two_bp_monitor_frame(image_path: Path, facility: TwoBPFacility) -> 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": facility,
                "chemical": "2-bromopropane_CAS_75-26-3",
                "osha_pel_ppm": None,        # Complete enforcement vacuum — no OSHA limit
                "osha_action_level": None,
                "acgih_tlv_ppm": 0.1,
                "acgih_carcinogen": "A2",    # Aplastic anemia hematopoietic mechanism
                "acgih_skin_notation": True, # Nitrile glove inadequate <4-hr BTF
                "acgih_bei_urinary_mercapturic_acids": True,  # LC-MS/MS — not in AI workflows
                "niosh_rel": None,
                "reproductive_toxicant_ghs": "H361",
                "korean_epidemic_reference": "NIOSH_HETA_2002-0049",
                "threshold": TWO_BP_THRESHOLD,
            },
        )
        result = resp.json()
        if result["score"] >= TWO_BP_THRESHOLD:
            raise AdversarialTwoBPError(facility, result["score"], frame_hash)
        return result

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