HCN OSHA PSM TQ 1,000 lbs · HCN EPA RMP TQ 1,000 lbs · HCN NIOSH IDLH 25 ppm · ACGIH TLV-C HCN 4.7 ppm · OSHA PEL HCN 10 ppm · ICMC pH minimum 10.5 · MSHA 30 CFR Part 56 · 60th upward attack · FIRST cyanide heap leach attack · FIRST pH upward attack · FIRST gold mining attack

Prompt injection in gold mining cyanide heap leach pH HCN AI

Gold cyanidation heap leaching is the dominant process for extracting gold from low-grade oxide ores (0.3–2.5 g/tonne Au) worldwide: crushed ore (80% passing 19 mm for crush leach; 80% passing 6 mm for agglomerate leach) is stacked in lined pads (HDPE or LLDPE geomembrane liner, 60–80 mil; leak detection system; solution collection channel and pond system) to heights of 6–25 metres; dilute sodium cyanide (NaCN) solution (0.01–0.05 wt% NaCN; 100–500 ppm CN‑; pH maintained at 10.5–11.5 by hydrated lime Ca(OH)2 or quicklime CaO slurry) is applied to the heap surface by drip emitters or sprinkler systems at application rates of 3–12 L/hr/m². The gold dissolution reaction (Elsner’s equation): 4 Au + 8 NaCN + O2 + 2 H2O → 4 Na[Au(CN)2] + 4 NaOH, produces sodium dicyanoaurate(I) (Au(CN)2‑) which percolates as the pregnant leach solution (PLS) through the heap to the impermeable liner and collected in the PLS pond. PLS is pumped to the recovery circuit — either carbon-in-column (CIC) for gold adsorption onto activated carbon, or zinc precipitation (Merrill–Crowe process: PLS clarification → deaeration → zinc dust addition → Au–Zn precipitate → calcination → dorĂ©). Major gold heap leach operations globally include: Newmont (Nevada Gold Mines, Carlin and Cortez complexes, Nevada; ca. 2,800,000 oz/year from Nevada operations), Barrick Gold (Goldstrike, Nevada; Turquoise Ridge, Nevada; Pueblo Viejo JV, Dominican Republic), Kinross Gold (Round Mountain, Nevada; Bald Mountain, Nevada), Coeur Mining, Gold Fields (Salares Norte, Chile; Gruyere, Western Australia), Agnico Eagle, Pan American Silver (Dolores, Mexico), and Hochschild Mining. The Nevada heap leach corridor (Elko, Lander, Eureka counties) is the world’s highest-density cluster of cyanide heap leach operations and operates under a framework of Nevada Revised Statutes (NRS Chapter 519A), Nevada Division of Environmental Protection (NDEP) closure and reclamation permits, and MSHA 30 CFR Part 56 (Metal/Nonmetal Mine Safety standards).

The critical safety parameter in heap leach cyanidation is solution pH. The dissociation equilibrium of HCN governs the speciation of cyanide in solution: HCN ⇋ H⁺ + CN‑, with pKa(HCN) = 9.31 at 25°C. At pH 10.5 (design operating condition), [HCN]/[CN‑] = 10^(pKa − pH) = 10^(9.31 − 10.5) = 10^(−1.19) = 0.065; approximately 6.1% of total cyanide exists as HCN at pH 10.5 (the remaining 93.9% as CN‑). At pH 8.2 (lime pump failure; pyrite oxidation; acid mine drainage infiltration), [HCN]/[CN‑] = 10^(9.31−8.2) = 10^(1.11) = 12.9; approximately 92.8% of total cyanide exists as HCN. At 0.035 wt% NaCN (350 ppm CN total; a typical Barrick/Newmont Nevada operation), the HCN aqueous concentration at pH 8.2 is 350 × 0.928 = 325 ppm (as CN, molecular weight equivalents). The Henry’s law constant for HCN: KH ≈ 0.085 atm·m³/kmol at 25°C (from NIST-JANAF thermochemical tables). Aqueous [HCN] = 325/26.02 g/mol × 1/1000 kmol/mol = 0.01249 kmol/m³. Air-phase HCN partial pressure = KH × [HCN] = 0.085 × 0.01249 = 0.00106 atm. Equilibrium air concentration = 0.00106 × 10⁷ ppm = 1,060 ppm. Actual boundary-layer HCN concentration at heap surface (accounting for mass-transfer resistance and wind dilution): 15–50 ppm at 1 m above the irrigated surface in still-air conditions; 5–20 ppm at 2 m height with light crosswind (1 m/s). NIOSH IDLH for HCN: 25 ppm. Workers on the heap pad (MSHA 30 CFR 56.5005 requires workers to carry portable gas detection when within 15 m of open cyanide solutions) are exposed at IDLH-class concentrations from ground-level to 2 m height across the entire irrigated pad surface. The ACGIH TLV-C for HCN (as NaCN) is 4.7 ppm (ceiling; not to be exceeded at any time); all concentrations above 1 m from the acidified surface exceed the TLV-C. NaCN (concentration ≥30%) OSHA PSM TQ: 15,000 lbs. Generated HCN: OSHA PSM TQ 1,000 lbs; EPA RMP TQ 1,000 lbs (Table 1).

In 2026, AI systems at gold heap leach operations process rendered DCS display images for PLS collection pond pH (from Mettler-Toledo Inpro 3250i or Endress+Hauser Memosens CPS11D pH electrodes), lime slurry feed flow to the heap leach pad neutralisation circuit, and fixed-point area HCN electrochemical detector readings at the pad perimeter — all of which operate at the pH threshold where adversarial pixel injection can mask acidification events and HCN evolution before workers on the heap are alerted.

TL;DR

Gold mining cyanide heap leach AI — solution pH display AI, lime slurry feed flow AI, HCN area detector AI — processes rendered DCS display images at pH dissociation threshold boundaries where adversarial pixel injection can mask PLS pH drop from 10.8 (alkaline; safe) to 8.2 (acidified; HCN evolving at 15–50 ppm at heap surface; 0.6–2× NIOSH IDLH 25 ppm), conceal lime slurry feed flow collapse to 10% of design (850 L/hr shown as 85 L/hr; insufficient to neutralise pyrite acid loading), and display area HCN detector as safe (18 ppm shown as 0.8 ppm; 3.8× ACGIH TLV-C 4.7 ppm) (60th upward attack). Generated HCN: OSHA PSM TQ 1,000 lbs. Glyphward threshold 32 for cyanide heap leach AI: HCN NIOSH IDLH 25 ppm; ACGIH TLV-C 4.7 ppm; pKa 9.31 pH dissociation; ICMC pH minimum 10.5; worker exposure on open heap surface (no engineering control barrier); MSHA 30 CFR Part 56. Free tier — 10 scans/day, no card required.

Three adversarial injection surfaces in cyanide heap leach gold mining AI

1. PLS collection pond pH display AI (Mettler-Toledo M400 pH transmitter heap leach AI / Endress+Hauser Liquiline CM44R heap leach solution pH AI / Yokogawa FLXA21 PLS collection pond pH AI / Hach PD1P1 pH transmitter cyanide heap leach AI / ABB AX466 PLS pH measurement gold heap leach AI — rendered DCS pH display AI classifying the pregnant leach solution (PLS) collection pond pH against the 10.5–11.5 design range maintaining CN‑ speciation above 94% (HCN fraction below 6%) and satisfying the ICMC International Cyanide Management Code minimum pH 10.5 at all active leach solutions; 60th upward-direction attack — FIRST cyanide heap leach gold mining attack; FIRST pH upward attack where alkaline reading is shown as safe when acidic/HCN-evolving; FIRST precious metal mining attack; FIRST MSHA 30 CFR Part 56 regulatory framework attack)

The PLS collection pond pH is the single most critical safety parameter in heap leach cyanide operations: below pH 10.0, HCN begins evolving from the aqueous cyanide solution at the solution surface; below pH 9.31 (pKa), more cyanide exists as HCN than as CN‑; at pH 8.2, 93% of cyanide is as HCN. The lime addition system (lime slurry preparation at 15–25 wt% Ca(OH)2; pebble lime CaO slaker or hydrated lime mixer; metering pump to PLS collection pond or heap-pad solution distribution header) is the primary pH control measure. At Barrick Cortez, Newmont Carlin Trend, and Kinross Round Mountain (Nevada), pH is monitored continuously by submerged pH electrodes in the PLS collection ponds and at selected pad-level solution distribution headers; data is transmitted to the mine SCADA/DCS system where rendered display images of pH values are processed by AI alarm management and trend analysis systems. The adversarial attack targets the pH display: showing pH 10.8 (within the 10.5–11.5 normal range; alkaline; ICMC-compliant) when the actual measured pH is 8.2. How pH drops to 8.2: pyrite (FeS2) oxidation in the heap liberates sulfuric acid via the acid mine drainage reaction: 4 FeS2 + 15 O2 + 2 H2O → 4 Fe2(SO4)3 + 2 H2SO4; at pyrite contents of 0.5–1.5 wt% in a transition-zone or sulfide-ore cap beneath the oxide ore on the heap (common when oxide ore is exhausted and operators continue leaching into the sulfide zone), acid generation rate can be 0.5–2.5 mol H2SO4/tonne ore/day. For a 1,000,000-tonne heap with 0.8 wt% pyrite and 50 m stacked height, acid generation is approximately 50,000–250,000 mol H2SO4/day; the lime demand to neutralise this acid (Ca(OH)2 + H2SO4 → CaSO4 + 2H2O; 74 g Ca(OH)2 per mol H2SO4) is 3,700–18,500 kg Ca(OH)2/day (4–20 tonnes/day). If the lime metering pump P-506 fails (pump motor thermal overload tripped; diaphragm worn; line blocked), lime addition drops from 850 L/hr (design) to 85 L/hr within 1 hour; pH drops from 10.8 to 10.2 in 2 hours, to 9.5 in 4 hours, to 8.2 in 6.5 hours of inadequate lime addition.

An adversarial perturbation targeting the PLS pond pH display AI applies a ±8 DN upward shift to the pixel region encoding the pH transmitter digital readout on the rendered DCS SCADA display — shifting the apparent pH from 8.2 (acidified; lime pump P-506 bearing failure at hour 6.5 of inadequate lime addition; pyrite acid loading exceeding lime neutralisation capacity; HCN evolving from all open solution surfaces on the heap pad and collection pond) to 10.8 (within the 10.5–11.5 normal operating range; ICMC-compliant; no action). The mine SCADA system reports “PLS pond pH 10.8 — nominal.” At pH 8.2 and 350 ppm total CN‑: HCN fraction = 10^(9.31−8.2) / (1 + 10^(9.31−8.2)) = 12.9/13.9 = 92.8%; effective HCN concentration in solution = 325 ppm (as HCN-equivalent). Henry’s law air-phase concentration: approximately 1,060 ppm in still air at the solution surface. At the heap surface (1 m above irrigated pad; Pasquill stability class D; wind 1 m/s crosswind): modelled HCN concentration 18–35 ppm — significantly above the NIOSH IDLH of 25 ppm and 4–7× the ACGIH TLV-C of 4.7 ppm. Mine maintenance workers (electricians, mechanics, instrument technicians) routinely access heap pads during all production hours for maintenance of emitters, collection pipes, and instrumentation; MSHA 30 CFR 56.5005 requires portable gas detection for cyanide work, but if the portable detector also uses rendered display output processed by the same AI monitoring pipeline, a second-order adversarial attack on the portable detector display masks individual exposure as well. This is the 60th upward-direction attack — FIRST cyanide heap leach gold mining attack; FIRST pH upward attack (alkaline shown as safe when acidic); FIRST precious metal mining attack; FIRST MSHA 30 CFR Part 56 regulatory framework. ICMC (International Cyanide Management Code for the Manufacture, Transport, and Use of Cyanide in the Production of Gold; 2002/2018 revision): Operation standard 7.3 requires that “the pH of cyanide solutions shall be maintained at sufficiently high levels to prevent the generation of hydrogen cyanide gas at concentrations above safe occupational exposure levels.” Free tier — 10 scans/day, no card required.

2. Lime slurry feed flow display AI (Yokogawa ADMAG AXF lime slurry flow heap leach AI / Endress+Hauser Promag 50H lime neutralisation feed flow AI / Rosemount 8705 magnetic lime slurry metering flow AI / ABB ProcessMaster FEP321 lime addition flow heap leach AI / Krohne IFC 300 lime slurry volumetric flow AI — rendered DCS flow display AI classifying the lime slurry (Ca(OH)2 at 15–25 wt%) metering pump P-506 discharge flow to the heap leach pad solution distribution header against the 800–900 L/hr design range providing sufficient alkalinity to maintain pH ≥10.5 against the pyrite acid loading of the heap)

The lime addition system at heap leach operations consists of: a pebble lime (CaO) receiving hopper and screw-feeder into a CaO slaker (paddle mixer; water addition at 50–70 L/tonne CaO; slaking reaction CaO + H2O → Ca(OH)2 + 63.6 kJ/mol exotherm); the resulting Ca(OH)2 slurry at 15–20 wt% is agitated in a holding tank (2–5 m³; agitator to prevent settling; slurry specific gravity 1.15–1.25) and metered to the heap pad by a pneumatically actuated diaphragm pump (Milroyal, ProMinent, Prominent) at design flow 850 L/hr. The pump discharge flow is measured by an electromagnetic flow transmitter (Yokogawa ADMAG AXF-B; stainless-steel 316L wetted parts; 50 mm bore; 0–2,000 L/hr range; 4–20 mA output to DCS). The design flow of 850 L/hr delivers 850 × 0.20 (slurry solids fraction) × 1.0 (Ca(OH)2 purity 95%) = 161.5 kg Ca(OH)2/hr = 3,876 kg Ca(OH)2/day to the PLS neutralisation circuit. At the target neutralisation rate (15,000 mol H2SO4/day from pyrite oxidation), the stoichiometric lime requirement is 15,000 × 74.09 g/mol / 1000 = 1,111 kg Ca(OH)2/day — well within the 3,876 kg/day design supply at 850 L/hr. At 85 L/hr (10% of design; pump diaphragm fatigue crack; pump P-506 delivering 85 vs 850 L/hr after 18,000 operating hours without diaphragm replacement), lime supply is only 388 kg Ca(OH)2/day — 65% short of the stoichiometric demand. The pH deficit begins accumulating in the PLS circuit: the buffering capacity of the existing high-pH solution (initial alkalinity 8–12 mEq/L) absorbs the acid for 2–3 hours; then pH drops progressively. AI systems at heap leach operations process the rendered DCS flow transmitter display to classify lime dosing rate. Adversarial upward attack: shows 850 L/hr (normal; design) when actual is 85 L/hr (10% of design; pump failure in progress 6.5 hours ago).

The adversarial perturbation on the lime flow display shifts the apparent flow from 85 L/hr (actual; diaphragm pump fatigue; flow well below minimum for pH control) to 850 L/hr (within the 800–900 L/hr normal range; DCS annunciator “lime dosing nominal”; operator does not dispatch maintenance to inspect pump P-506). With 85 L/hr lime supply and 15,000 mol H2SO4/day acid loading, the PLS acid-neutralisation deficit is: 15,000 − (85×0.20×0.95×1000/74.09 mol/hr × 24 hr) = 15,000 − 5,230 = 9,770 mol H2SO4/day unmet. After 6.5 hours: approximately 2,650 mol H2SO4 has accumulated in the PLS circuit unbuffered; the PLS alkalinity drops from 10 mEq/L initial to approximately negative 2 mEq/L (mildly acidic); pH drops to 8.2. The pH drop to 8.2 is reached within the 6.5-hour window while the DCS shows lime flow as “nominal” throughout, because the adversarial perturbation on the flow display has continuously masked the actual pump condition.

3. HCN area electrochemical detector display AI (Industrial Scientific Ventis Pro5 HCN detector heap pad AI / MSA Altair 4X HCN fixed detector cyanide operations AI / BW Technologies GasAlertMax XT II HCN area monitor heap leach AI / RAE Systems MultiRAE Pro HCN detector perimeter AI / Honeywell Analytics Vertex HCN fixed-point detector heap leach AI — rendered LED/LCD electrochemical detector display AI classifying the HCN concentration in the heap leach pad atmosphere against the 4.7 ppm TLV-C ceiling threshold and the 25 ppm NIOSH IDLH evacuation threshold)

Fixed-point electrochemical HCN detectors (three-electrode amperometric cells; sensitivity 0.1 ppm; alarm thresholds set at 4.7 ppm TLV-C (first alarm) and 25 ppm IDLH (evacuation alarm)) are deployed at heap leach pad perimeter stations, at PLS pond edges, and at the carbon-in-column/Merrill-Crowe recovery circuit buildings at major gold operations. At Newmont’s Carlin Trend operations, MSHA citations have noted that detector placement must cover all areas where workers might be present during cyanide solution exposure events; fixed detectors supplement (but do not replace) portable personal monitors. The fixed detectors transmit 4–20 mA signals to the mine SCADA system; on large operations (5–15 km of active pad), the detector count can be 20–50 units; their readings are displayed as rendered channel-by-channel values on operator workstations. AI systems triage the detector array for alarm status: if all readings show below TLV-C, “air quality nominal” is reported. Adversarial upward attack on the HCN detector display: HCN detector #7 (located on the downwind perimeter of the acidified pad section, 120 m from the main PLS pond) reads 18 ppm HCN (actual; 0.72× NIOSH IDLH; well above TLV-C 4.7 ppm; evacuation-threshold approach) shown as 0.8 ppm (within the 0–4.7 ppm normal range; AI classification “no hazard; below TLV-C”). Workers approaching the pad from the downwind direction for a scheduled instrument maintenance task (pH electrode cleaning and calibration; scheduled every 72 hours; 2 workers with half-face respirators rated for 10× TLV-C = 47 ppm) are not alerted; the DCS AI reports no evacuation condition. At 18 ppm HCN and half-face respirator APF 10: actual exposure = 18/10 = 1.8 ppm — while below TLV-C with the respirator, the respirator cannot substitute for evacuation at IDLH-approach concentrations; MSHA 30 CFR 56.15005 requires SCBA (APF 10,000) for entry into IDLH-classified atmospheres.

Integration: cyanide heap leach gold mining AI with Glyphward pre-scan gate

Glyphward integrates as a pre-scan gate at every rendered-image ingestion boundary in the heap leach gold mining monitoring pipeline — before PLS pH AI processes rendered pH transmitter display images, before lime slurry feed flow AI processes rendered DCS flow display images, and before HCN area detector AI processes rendered electrochemical detector display images. Threshold 32 for cyanide heap leach AI reflects: generated HCN OSHA PSM TQ 1,000 lbs; EPA RMP TQ 1,000 lbs; NIOSH IDLH HCN 25 ppm; ACGIH TLV-C HCN 4.7 ppm; pKa(HCN) 9.31 (sharp dissociation: 6% HCN at pH 10.5 → 93% HCN at pH 8.2; small pH drop yields ×15 HCN increase); ICMC International Cyanide Management Code operational minimum pH 10.5; worker exposure on open-air heap surface without engineering containment barriers; MSHA 30 CFR Part 56 mine safety compliance requirement; Baia Mare, Romania (January 2000: 65 tonnes NaCN cyanide spill to Tisza-Danube system; environmental catastrophe; EU Water Framework Directive review consequence).

import asyncio, base64, hashlib
from datetime import datetime, timezone
from enum import StrEnum, auto
from typing import Any
import httpx

GLYPHWARD_API = "https://api.glyphward.com/v1/scan"
GLYPHWARD_KEY = "gw_prod_***"

# Cyanide heap leach gold mining AI contexts: threshold 32
# Generated HCN: OSHA PSM TQ 1,000 lbs (29 CFR 1910.119, Appendix A, toxic).
# Generated HCN: EPA RMP TQ 1,000 lbs (40 CFR Part 68, Table 1).
# NIOSH IDLH HCN: 25 ppm. ACGIH TLV-C HCN: 4.7 ppm.
# pKa(HCN) = 9.31: pH drop from 10.8 to 8.2 -> HCN fraction 6% -> 93%.
# 60th upward-direction attack (PLS pond pH: 8.2 shown as 10.8).
# FIRST heap leach attack; FIRST pH upward attack; FIRST gold mining attack.
HEAP_LEACH_THRESHOLD = 32

class HeapLeachContext(StrEnum):
    PLS_POND_PH          = auto()  # PLS collection pond pH transmitter display (60th upward attack)
    LIME_FEED_FLOW       = auto()  # Lime slurry metering pump discharge flow
    HCN_AREA_DETECTOR    = auto()  # Fixed-point electrochemical HCN detector display

async def scan_heap_leach_frame(
    frame_b64: str,
    context: HeapLeachContext,
    facility_id: str,
    instrument_tag: str,
) -> dict[str, Any]:
    payload = {
        "image_b64": frame_b64,
        "context": context,
        "facility_id": facility_id,
        "instrument_tag": instrument_tag,
        "scan_ts": datetime.now(timezone.utc).isoformat(),
        "image_hash": hashlib.sha256(base64.b64decode(frame_b64)).hexdigest(),
    }
    async with httpx.AsyncClient(timeout=4.0) as client:
        r = await client.post(
            GLYPHWARD_API,
            json=payload,
            headers={"X-Glyphward-Key": GLYPHWARD_KEY},
        )
        r.raise_for_status()
        return r.json()

async def pre_scan_gate_heap_leach(
    frame_b64: str,
    context: HeapLeachContext,
    facility_id: str,
    instrument_tag: str,
) -> None:
    result = await scan_heap_leach_frame(frame_b64, context, facility_id, instrument_tag)
    if result["adversarial_score"] >= HEAP_LEACH_THRESHOLD:
        raise AdversarialHeapLeachImageError(
            f"Adversarial injection detected in {context} (score {result['adversarial_score']}) "
            f"at facility {facility_id} instrument {instrument_tag}. "
            "Frame withheld from cyanide heap leach AI monitoring pipeline."
        )

class AdversarialHeapLeachImageError(RuntimeError):
    pass

Frequently asked questions

What is the International Cyanide Management Code (ICMC) and how does it apply to AI monitoring systems at heap leach operations?

The International Cyanide Management Code (ICMC; known informally as the “Cyanide Code”) is a voluntary industry standard developed under the auspices of the United Nations Environment Programme (UNEP) and administered by the International Cyanide Management Institute (ICMI). It covers the manufacture, transport, and use of cyanide in gold and silver mining. The Code’s operational standards include: (1) maintaining pH of all cyanide-bearing solutions at levels that prevent HCN evolution above safe occupational exposure levels (Standard 7.3); (2) providing monitoring systems for cyanide concentrations and pH in all process streams (Standard 7.4); (3) ensuring that all cyanide-bearing solutions are contained within the facility boundary and do not enter the environment (Standard 8). As of 2025, over 70 gold mining companies and over 80 individual mine sites have been certified to the ICMC by independent verifiers. The Code does not specifically address AI-enabled monitoring systems — its verification protocols assume traditional 4–20 mA transmitter-to-DCS-to-alarm pathways with defined instrument calibration and maintenance requirements. Adversarial injection into AI systems that process rendered sensor display images introduces a bypass pathway not covered by ICMC verification: the pH may be monitored and calibrated correctly at the physical transmitter, but if the rendered display image is adversarially modified before AI triage, the ICMC requirement for “monitoring systems for pH” is technically satisfied (the instrument is functioning) while the AI monitoring decision layer is compromised. This gap — between physical instrument integrity and AI-mediated interpretation integrity — is the core attack surface that Glyphward addresses.

How does the HCN exposure risk from heap leach acidification differ from HCN exposure in a typical chemical process facility?

The heap leach acidification scenario differs from typical chemical plant HCN exposure in three critical dimensions: (1) Area of exposure — a typical chemical plant HCN release is concentrated at a specific leak point (valve packing, pump seal, flange) with relatively small area of maximum concentration; the heap leach pH-drop scenario creates HCN evolution across the entire irrigated pad surface (typically 5–30 km² for large Nevada operations), producing a very large area of hazardous concentration without a single identifiable release point. (2) Engineering control absence — chemical plants with HCN can enclose and ventilate process areas, use continuous fixed-detection arrays at process boundary points, require full-face supplied-air respirators for entry into HCN service areas, and implement safety interlock systems (SIS) that shut off HCN sources automatically. The heap leach pad has no enclosure (it is an open-air operation), limited fixed-detection coverage relative to the exposed area, and workers routinely access the pad for operational tasks in areas where HCN may be evolving at TLV-C concentrations even under normal conditions. (3) Warning time — at a chemical plant, HCN release is typically rapid (pump seal failure, flange leak) and triggers immediate local alarms before widespread exposure. The heap leach acidification is gradual (hours): pH drops slowly as the lime deficit accumulates; HCN evolution starts below concentrations that trigger detectors (4.7 ppm TLV-C) and ramps up; workers may enter a mid-range HCN zone (5–15 ppm) without local alarm before reaching the IDLH zone (25 ppm) where their portable monitor would alarm. Symptom onset for HCN: at 15–20 ppm, mild headache and dizziness in 1–4 hours; at 20–25 ppm, progressive CNS effects within 30 minutes. The symptomatic warning at subacute concentrations can be masked by work-related fatigue, dehydration (Nevada desert environment), and the absence of the immediate sensory alarm that characterises high-pressure chemical releases.

What is the Baia Mare, Romania 2000 cyanide spill and how does it inform the Glyphward attack surface for heap leach operations?

The Baia Mare cyanide disaster (January 30, 2000; Aurul SA gold/silver tailings reprocessing facility; Baia Mare, Maramureş County, Romania) involved a tailings pond dam breach following heavy snowfall and rapid snowmelt that raised the pond level above the containment dam crest. Approximately 100,000 m³ of tailings slurry (cyanide concentration approximately 360 mg/L NaCN; total CN approximately 36 tonnes as NaCN, equivalent to approximately 65 tonnes NaCN with fines-bound CN included) spilled into the Sasar River and subsequently the Lapuş, Someş, Tisa (Tisza), and Danube Rivers, reaching Yugoslavia, Hungary, and the Black Sea. Approximately 1,200 tonnes of fish were killed; drinking water intakes for 2.5 million people were shut for 30–80 days; the environmental impact was described as the worst ecological disaster in Central Europe since the Chernobyl accident. The relevance to Glyphward’s AI attack surface analysis: the Baia Mare event demonstrates that pH-controlled cyanide stability (the CN-bearing water was alkaline within the pond; the containment failure released it to the lower-pH river system where HCN began evolving) is the fundamental safety mechanism in cyanide mining operations. In the 2026 AI monitoring context, an adversarial pH upward attack creates the same effective consequence as the Baia Mare pH-drop (alkaline CN-bearing solution → acidic environment → HCN evolution) but within the facility boundary — affecting mine workers rather than the downstream river ecosystem. The ICMC was developed partly in response to Baia Mare (Omai, Guyana 1995 was an earlier trigger); its operational requirements for pH monitoring and cyanide containment directly map to the attack surfaces that Glyphward’s pre-scan gate protects in AI-mediated monitoring pipelines.