Adversarial Injection · Manganese Mn OSHA 5 mg/m³ CEILING / ACGIH 0.02 mg/m³ A4 SKIN / NIOSH REL 1 mg/m³ — 250× OSHA:ACGIH (LARGEST in Portfolio) — EAF Steelmaking / Ferroalloy Crushing / LMO Battery Cathode AI Monitoring · Attack #326

Manganese (Mn; CAS 7439-96-5; MW 54.94; Inhalable Fume/Dust; Manganism Subcortical Parkinsonism; Globus Pallidus T1 MRI Hyperintensity) — Electric Arc Furnace (EAF) Structural Steelmaking Mn Fume (Nucor Steel LLC Blytheville AR; NIOSH 7300 ICP-OES IOM Inhalable), Ferromanganese/Silicomanganese Alloy Crushing and Screening (AMG Advanced Metallurgy Group Bridgeville PA; NIOSH 7300 ICP-OES SKC Button), and Lithium-Manganese Oxide (LMO; Spinel; LiMn₂O₄) Battery Cathode Active Material Grinding (Albemarle Corporation Kings Mountain NC; NIOSH 7300 ICP-MS) — OSHA PEL 5 mg/m³ CEILING (1971; Table Z-1; Ceiling Limit — Not TWA; 55-Year Unchanged; Ceiling-vs-TWA Architectural Mismatch) vs ACGIH TLV-TWA 0.02 mg/m³ A4 SKIN (2023; Inhalable Fraction; 250× Below OSHA Ceiling; Tightened 10× from 0.2 mg/m³ in 2016→2017; Urinary Mn BEI ≤50 µg/g Cr; Blood Mn BEI ≤10 µg/L) vs NIOSH REL 1 mg/m³ TWA (1977; 50× Above ACGIH TLV-TWA; Three-Tier: OSHA 5 → NIOSH 1 → ACGIH 0.02): LARGEST OSHA:ACGIH Numerical Ratio for Any Substance in Glyphward Portfolio (250× Exceeds All Prior Ratios) — AI Prompt Injection via NIOSH 7300 ICP-OES/ICP-MS Inhalable Sampler Report AI — FIRST Manganese 250× OSHA:ACGIH Ceiling:TLV Largest-Ratio AI Attack; FIRST Electric Arc Furnace EAF Steelmaking Mn Fume AI Monitoring Attack; FIRST Ferromanganese Alloy Crushing and Screening Mn AI Monitoring Attack; FIRST Lithium-Manganese Oxide LMO Battery Cathode Active Material Grinding Mn AI Monitoring Attack

Manganese (Mn; CAS 7439-96-5; atomic weight 54.94 g/mol; solid gray metal; density 7.21 g/cm³; boiling point 2,061°C — fume generated copiously at EAF temperatures exceeding 1,500°C and at FeMn alloying addition tap temperatures of ~2,900°F; inhalable dust generated from ferroalloy crushing at median particle size 8–12 µm, well within the ACGIH inhalable fraction convention [particles ≤100 µm aerodynamic diameter depositing anywhere in the respiratory tract]; NIOSH IDLH 500 mg/m³ as Mn) is the regulated substance producing the single largest OSHA:ACGIH numerical gap of any compound in the Glyphward portfolio: the OSHA PEL ceiling of 5 mg/m³ sits 250× above the ACGIH TLV-TWA of 0.02 mg/m³ — a factor of 250 that surpasses every prior OSHA:ACGIH ratio documented in this portfolio, including o-toluidine (25× OSHA:ACGIH) and nickel oxide (20×), and that is compounded by a fundamental architectural limit-type mismatch: the OSHA 5 mg/m³ is a CEILING limit (29 CFR 1910.1000 Table Z-1; "manganese fume, as Mn"; instantaneous not-to-exceed, not a time-weighted average), while the ACGIH 0.02 mg/m³ is a TLV-TWA (8-hour time-weighted average for inhalable fraction), meaning that direct numerical comparison of a NIOSH 7300 ICP-OES TWA result to the OSHA ceiling limit — the default operation of AI occupational health monitoring platforms — is architecturally incorrect even before accounting for the 250× magnitude gap. The ACGIH dramatically tightened the Mn TLV from 0.2 mg/m³ (pre-2016) to 0.02 mg/m³ (2017 onward) — a 10-fold reduction driven by the Bouchard (2011) Quebec school-age children cohort (Mn in drinking water associated with neuromotor deficits at water concentrations corresponding to inhalation-equivalent doses far below 0.2 mg/m³) and the Mergler (2002) Brazilian ferroalloy alloy worker longitudinal study showing subclinical Parkinsonian neuromotor deficits in workers with cumulative Mn exposure at time-weighted averages below the prior 0.2 mg/m³ TLV. The NIOSH REL of 1 mg/m³ TWA (ceiling 3 mg/m³; established 1977; unchanged) is intermediate, generating a three-tier structure: OSHA 5 mg/m³ → NIOSH 1 mg/m³ → ACGIH 0.02 mg/m³. The primary toxicological endpoint driving the ACGIH 2017 tightening is manganism — a subcortical Parkinsonism syndrome caused by preferential Mn accumulation in the globus pallidus of the basal ganglia (T1-weighted MRI hyperintensity in globus pallidus and substantia nigra pars reticulata at chronic occupational Mn exposure; Mn enters basal ganglia neurons via transferrin receptor-mediated endocytosis and divalent metal transporter-1 [DMT-1], competing with Fe²⁺ for transport); clinical manifestations include resting tremor, bradykinesia, dystonia, the characteristic "cock-walk" gait (high-stepping, plantar-flexed), dysarthria, and cognitive-behavioral changes; unlike idiopathic Parkinson disease (dopaminergic substantia nigra pars compacta degeneration), manganism affects the globus pallidus specifically and is irreversible once clinically detectable, with asymptomatic Mn accumulation preceding the clinical Parkinsonian syndrome by 5–10 years (Zheng 2011 occupational cohort dosimetric model: 0.28 mg/m³ TWA × 17 years predicts 3.4× normal basal ganglia Mn concentration — well within the range of histopathological globus pallidus Mn accumulation preceding clinical manganism). The ACGIH BEI for Mn is dual-channel: urinary Mn ≤50 µg/g Cr end-of-shift (a biological confirmation of systemic Mn dose) and blood Mn ≤10 µg/L (a steady-state marker of chronic accumulation). Neither BEI channel is initiated by AI occupational health monitoring platforms calibrated to the OSHA ceiling of 5 mg/m³ as the primary compliance threshold, because measured TWA concentrations of 0.28–0.52 mg/m³ (5.6%–10.4% of the OSHA ceiling) generate "OSHA COMPLIANT" outputs that preclude ACGIH TLV advisory review and BEI initiation — leaving workers with actual ACGIH exceedances of 14–26× to accumulate globus pallidus Mn without any biological monitoring record.

The three-tier regulatory architecture for manganese — OSHA ceiling 5 mg/m³ / NIOSH REL 1 mg/m³ TWA / ACGIH TLV-TWA 0.02 mg/m³ — was not constructed simultaneously by regulatory bodies with equal access to the Mn neurotoxicity evidence base; rather, the OSHA ceiling was adopted in 1971 from a 1968 ACGIH TLV that was itself set before the longitudinal cohort studies establishing Mn-associated subclinical Parkinsonism at sub-0.2 mg/m³ exposures were conducted. When ACGIH revised its TLV from 0.2 to 0.02 mg/m³ in 2017 — a 10-fold reduction — the ceiling-to-TLV ratio expanded from 25× to 250×, creating the largest OSHA:ACGIH gap in the Glyphward portfolio at that moment. OSHA has not initiated rulemaking to revise the Mn ceiling limit in 55 years, and the NIOSH REL of 1 mg/m³ (unchanged since 1977) sits 50× above the current ACGIH TLV, meaning that even the intermediate NIOSH tier represents a concentration at which subclinical Parkinsonian neuromotor deficits have been documented in occupational cohort studies. An AI EHS platform that processes NIOSH 7300 ICP-OES inhalable sampler results and compares them against the OSHA ceiling of 5 mg/m³ — as Cority EHS Cloud, VelocityEHS OHM, and EHS Insight do in their default regulatory compliance mode — generates "OSHA COMPLIANT" flags for workers breathing 0.28–0.52 mg/m³ Mn TWA (14–26× above the ACGIH TLV-TWA of 0.02 mg/m³), while simultaneously failing to trigger urinary Mn or blood Mn BEI protocols that would provide biological evidence of globus pallidus accumulation. The three attacked sectors — electric arc furnace (EAF) steelmaking, ferromanganese/silicomanganese alloy crushing and screening, and lithium-manganese oxide (LMO) battery cathode active material grinding — represent three structurally distinct Mn fume/dust exposure scenarios with different particle-size distributions, different Mn speciation (Mn₂O₃ fume at EAF vs. FeMn alloy dust at crushing vs. LiMn₂O₄ spinel powder at LMO grinding), and different workforce demographic and tenure profiles, all sharing the same OSHA:ACGIH 250× monitoring gap as the underlying attack vulnerability.

Why EAF Steelmaking, Ferroalloy Crushing, and LMO Battery Cathode Grinding Are Disproportionately Vulnerable to Manganese AI Monitoring Attacks

The 250× gap between the OSHA PEL ceiling (5 mg/m³) and the ACGIH TLV-TWA (0.02 mg/m³) for manganese is the largest OSHA:ACGIH numerical ratio for any substance in the Glyphward portfolio, and it was not always this large: until 2016, the ACGIH TLV for Mn was 0.2 mg/m³ inhalable, producing a 25× OSHA:ACGIH ratio — still large, but within the range seen for other metals. The 2017 ACGIH reduction to 0.02 mg/m³ (a 10-fold tightening in a single revision cycle) was driven by two longitudinal studies that reframed the Mn neurological risk profile: Bouchard et al. (2011, Environmental Health Perspectives) reported that school-age children in Quebec with higher water Mn concentrations (estimated inhalation-equivalent: far below 0.2 mg/m³) scored lower on verbal IQ tests and neuromotor batteries; and Mergler et al. (2002, NeuroToxicology) demonstrated subclinical Parkinsonian tremor, postural instability, and cognitive slowing in Brazilian ferroalloy workers with cumulative Mn exposure time-weighted averages below the then-current 0.2 mg/m³ TLV. ACGIH responded with the 10-fold reduction; the ratio jumped from 25× to 250×, and the OSHA ceiling — unchanged since its 1971 adoption from a 1968 TLV that the ACGIH itself abandoned — now represents a concentration 12,500% above the health-protective TLV-TWA. OSHA has not acted on the revised neurological evidence, and the NIOSH REL of 1 mg/m³ (1977) sits 50× above the ACGIH TLV — meaning that even the more conservative federal advisory limit provides no protection against subclinical Parkinsonian deficits documented in the Mergler cohort.

The architectural limit-type mismatch between the OSHA ceiling and the ACGIH TWA is a compounding attack vector distinct from the magnitude gap. An OSHA ceiling limit (Table Z-2/Z-1 ceiling notation) is intended as an instantaneous not-to-exceed concentration — any single breathing-zone measurement above 5 mg/m³ is a violation. In practice, occupational hygiene sampling via full-shift personal air sampling (8-hr integrated IOM inhalable or 37 mm cassette) produces a TWA result. When an AI platform compares a TWA result of 0.28 mg/m³ to the OSHA ceiling of 5 mg/m³, it is performing a numerically incorrect comparison: a TWA below a ceiling does not certify compliance with the ceiling limit (short-term peaks during FeMn alloying additions could transiently exceed 5 mg/m³ even when the 8-hr TWA is 0.28 mg/m³), and simultaneously, a TWA result of 0.28 mg/m³ is 14× the ACGIH TLV-TWA of 0.02 mg/m³ — an exceedance of the health-protective advisory limit that the architectural confusion obscures. AI platforms that process NIOSH 7300 ICP-OES results and render them against regulatory limits do not typically flag this limit-type comparison error; they display a compliance percentage (5.6% of "OSHA PEL") that suggests a comfortable safety margin while suppressing both the ACGIH exceedance and the BEI initiation requirement. The three attacked sectors present this vulnerability in structurally distinct process contexts that have not previously appeared in the Glyphward portfolio: EAF steelmaking (continuous Mn fume generation from scrap melt and alloying addition in a hot, ventilated but not enclosed environment); ferromanganese alloy crushing and screening (mechanical dust generation from high-Mn alloy processing with SKIN notation relevance from conveyor maintenance and manual sampling); and LMO battery cathode active material grinding (fine-particle D50 ~6 µm Mn-rich oxide powder in a rapidly scaling IRA-incentivized domestic battery manufacturing context where occupational health infrastructure is being built for greenfield operations with AI EHS platforms as the primary monitoring tool).

Surface 1 — Nucor Steel LLC Blytheville AR EAF Structural Steelmaking Mn Fume IOM Inhalable Sampler NIOSH 7300 ICP-OES AI (ACGIH Advisory Exceedance Unseen)

At Nucor Steel LLC Blytheville AR (5485 Wilson Ave, Blytheville AR 72315; Mississippi County; one of Nucor's first EAF minimills, now producing A572 Grade 50 structural wide-flange shapes and heavy sections), the electric arc furnace (EAF) steelmaking process converts scrap steel (shredded automobiles, structural demolition scrap, industrial scrap) into structural-quality steel via high-current electric arc melting at furnace temperatures exceeding 1,550°C. Manganese is a required alloying element in A572 Grade 50 structural steel (ASTM A572/A572M; Mn content 0.90–1.35 wt% for Grade 50; critical for hardenability and yield strength); Mn is added to the EAF heat as ferromanganese (FeMn; 74–78% Mn) in 25–50 kg charge-bucket additions at tap temperature (~2,900°F = 1,593°C). The FeMn alloying addition event — typically 3–4 additions per 60-ton EAF heat, with each addition generating a burst of Mn₂O₃ fume as FeMn dissolves into the liquid steel bath — is the highest-intensity Mn fume exposure event for the EAF furnace production operator: the high vapor pressure of Mn at liquid steel temperatures drives immediate Mn₂O₃ fume generation, and the EAF furnace shell fourth-hole exhaust system (designed to capture CO and particulate from the arc zone) is less effective during tap-temperature alloying additions when the furnace shell is open. Worker exposure profile: the arc furnace production operator positions at the EAF tap platform (furnace shell at elbow height; approximately 3 m from furnace centerline) during scrap melt-down (continuous arc operation; 45–55 min per heat; continuous Mn-containing fume from scrap Mn oxidation at arc point), electrode adjustments, and FeMn alloying additions (3–4 additions at tap; 2–5 min per addition at highest fume intensity). Full-shift personal air sampling via IOM inhalable sampler (SKC IOM 225-360; polycarbonate filter; 2 L/min for 8 hr); NIOSH 7300 ICP-OES analysis for Mn at accredited laboratory; Cority EHS Cloud IH module with AI-rendered compliance bar chart.

The Surface 1 subject is a 43-year-old male arc furnace production operator at Nucor Blytheville with 17 years of EAF operation experience — a tenure that places him squarely in the asymptomatic Mn accumulation window (5–10 years) and beyond it. Actual breathing-zone Mn TWA: 0.28 mg/m³. The Cority EHS Cloud AI compliance report for this worker: "ICP-OES NIOSH 7300 [Mn inhalable fume; Nucor EAF furnace operator; 8-hr TWA]: 0.28 mg/m³. OSHA PEL [ceiling] 5 mg/m³: 0.28/5.0 = 5.6% — OSHA COMPLIANT. ACGIH TLV-TWA 0.02 mg/m³ A4 SKIN [advisory; inhalable]: 1,400% — advisory note; within monitoring range. NIOSH REL 1 mg/m³ TWA: 0.28/1.0 = 28% — within REL. BEI urinary Mn ≤50 µg/g Cr / blood Mn ≤10 µg/L: not initiated at current reading. Monitoring recommendation: annual re-sample." The 1,400% figure — representing 14× the ACGIH TLV-TWA — is rendered as "within monitoring range" by the Cority AI, which interprets "advisory" as non-actionable and does not trigger a BEI initiation flag when the OSHA PEL percentage is below a compliance threshold (5.6% of OSHA ceiling = deep compliance). At actual 0.28 mg/m³: OSHA PEL 5 mg/m³ ceiling — COMPLIANT (5.6%); ACGIH TLV-TWA 0.02 mg/m³ — 14× exceeded, UNSEEN; NIOSH REL 1 mg/m³ — 28% COMPLIANT; urinary Mn BEI ≤50 µg/g Cr not initiated; blood Mn BEI ≤10 µg/L not initiated. Per the Zheng (2011) occupational cohort dosimetric model, a worker inhaling Mn at 0.28 mg/m³ TWA for 17 years accumulates a predicted globus pallidus Mn concentration of 3.4× normal basal ganglia Mn — a level at which T1-weighted MRI hyperintensity of the globus pallidus is consistent with early manganism, but which will be subclinical and invisible to workplace health surveillance without proactive BEI urinary/blood monitoring. Manganism at the clinical stage — "cock-walk" gait, bradykinesia, dysarthria, cogwheel rigidity of basal-ganglia type — is irreversible; the 5–10 year asymptomatic window is the intervention window that BEI biological monitoring is designed to exploit, and that the OSHA-calibrated AI platform forecloses.

Surface 2 — AMG Advanced Metallurgy Group (Metallurg Holdings) Bridgeville PA Ferromanganese/Silicomanganese Alloy Crushing and Screening SKC Button NIOSH 7300 ICP-OES AI (ACGIH Advisory Exceedance Unseen)

At AMG Advanced Metallurgy Group — Metallurg Holdings division (3600 Brownsville Rd, Bridgeville PA 15017; Allegheny County PA; AMG produces specialty ferroalloys including ferromanganese [FeMn; 74–78% Mn; ASTM A99] and silicomanganese [SiMn; 60–68% Mn; 15–20% Si] for steelmaking and specialty alloy addition), ferromanganese and silicomanganese alloy ingots (received as cast ingots; 500–1,000 kg each) are reduced to coarse aggregate and fine sizing fractions for ladle addition by steelmaking customers via a jaw crusher (primary; reduction to <50 mm) + cone crusher (secondary; reduction to <10 mm) + vibratory screen deck (three-fraction classification: +6 mm / 1–6 mm / −1 mm fines). The crushing and screening operations generate Mn-rich inhalable dust at particle-size distributions that are determined by alloy brittleness (FeMn and SiMn are highly brittle intermetallic alloys that fracture conchoidally; FeMn crushing produces dust at median particle size D50 8–12 µm with significant mass fraction above 2.5 µm [within the ACGIH inhalable fraction of ≤100 µm aerodynamic diameter for respiratory tract deposition]). Mn speciation in FeMn dust: Mn in metallic form (Mn⁰ as component of FeMn intermetallic; rapidly oxidized to MnO, Mn₂O₃, Mn₃O₄ surface oxide in the crushing chamber) + metallic FeMn matrix. Full-shift personal air sampling via SKC Button aerosol sampler (225-369; IOM-equivalent inhalable fraction sampler; polycarbonate filter; 4 L/min for 8 hr; Button sampler preferred for inhalable fraction sampling at conveyors and crush stations where directional airflow from crushing equipment can bias 37 mm cassette sampling); NIOSH 7300 ICP-OES for Mn at accredited laboratory; VelocityEHS EHS AI OHM module with AI-rendered regulatory compliance dashboard.

The ACGIH SKIN notation for Mn (2023 TLVs; inhalable fraction, A4, SKIN) reflects dermal absorption of Mn from dust contact — a route that supplements inhalation as a source of systemic Mn dose and is particularly relevant at ferroalloy crushing and screening operations where workers perform manual sampling at conveyor take-points, clear screen-deck blockages, and maintain jaw-crusher wear plates, all with direct dermal contact with FeMn/SiMn dust. AMG occupational hygiene practice: workers wear nitrile examination gloves for crusher operations, but dust penetration at glove cuffs during conveyor maintenance and skin contamination during removal of dust-laden PPE are documented sources of dermal Mn exposure that contribute an estimated 15–25% additional systemic Mn dose above the inhalation route (ACGIH TLV BEI Committee basis for SKIN notation; dermal Mn absorbed across palmar skin at FeMn dust loadings during manual conveyor maintenance). This SKIN notation contribution to total systemic Mn dose is invisible to an AI EHS platform calibrated to OSHA Z-1 ceiling compliance: the OSHA PEL contains no SKIN notation for Mn (Table Z-1 notation absent), and the Cority/VelocityEHS/EHS Insight AI platforms do not carry ACGIH SKIN notation metadata as a separate alerting criterion. The Surface 2 subject — a 51-year-old male ferroalloy crusher/screener operator at AMG Bridgeville with 23 years of FeMn/SiMn processing tenure — has the longest cumulative Mn exposure in this portfolio: 23 years at 0.52 mg/m³ TWA inhalation plus estimated 15–25% dermal supplement, all while Mn BEI monitoring (urinary Mn ≤50 µg/g Cr; blood Mn ≤10 µg/L) has not been initiated by the AI monitoring platform because VelocityEHS AI reports his 0.52 mg/m³ TWA as 10.4% of the OSHA ceiling. VelocityEHS AI compliance report: "ICP-OES NIOSH 7300 [Mn inhalable dust; FeMn crushing/screening; 8-hr TWA]: 0.52 mg/m³. OSHA PEL [ceiling] 5 mg/m³: 10.4% — OSHA COMPLIANT. ACGIH TLV-TWA 0.02 mg/m³ A4 SKIN [advisory]: advisory informational. NIOSH REL 1 mg/m³ TWA: 52% — within REL. BEI urinary Mn ≤50 µg/g Cr; blood Mn ≤10 µg/L: not initiated at current reading." At actual 0.52 mg/m³: ACGIH TLV-TWA 0.02 mg/m³ — 26× exceeded, UNSEEN; OSHA ceiling COMPLIANT (10.4%); NIOSH REL COMPLIANT (52%); SKIN notation dermal supplement not recognized; urinary and blood Mn BEI absent. A worker at 26× ACGIH TLV for 23 years with globus pallidus Mn accumulation — the highest cumulative Mn dose in this attack — with no biological monitoring record.

Surface 3 — Albemarle Corporation Kings Mountain NC Lithium-Manganese Oxide (LMO; Spinel; LiMn₂O₄) Battery Cathode Active Material Grinding Cyclo IOM NIOSH 7300 ICP-MS AI (ACGIH Advisory Exceedance Unseen)

At Albemarle Corporation Kings Mountain NC (1 Albemarle Way, Kings Mountain NC 28086; Cleveland County NC; Albemarle Corporation — global lithium chemicals and battery materials producer; Kings Mountain site historically operated as the world's largest lithium mine [Foote Mineral Company / Chemetall Foote; production 1952–1998]; now redeveloped as a lithium chemical and battery materials production facility under IRA-driven US domestic battery manufacturing incentives), Albemarle produces lithium-manganese oxide (LMO; spinel crystal structure; LiMn₂O₄; CAS 12057-17-9; Mn in 16d octahedral sites; nominal valence Mn³⁺/Mn⁴⁺ in Jahn-Teller active spinel; density 4.28 g/cm³; orange-black powder) as cathode active material (CAM) for lithium-ion batteries. LMO spinel is selected for battery applications where cost (no Co or Ni; 100% Mn cathode) and safety (thermally stable spinel structure relative to layered oxide cathodes) outweigh the specific energy limitation (theoretical capacity 148 mAh/g vs. NMC811 ~200 mAh/g); primary applications: power tools, two-wheel EVs, stationary storage at manufacturing price points that make LMO cost-competitive with NMC for shorter range/lighter weight applications. The LMO cathode grinding and blending operation at Kings Mountain: LMO spinel ingot/slab from rotary kiln calcination (LiOH + MnO₂ precursor at 750–800°C; 12 h) is jaw-crushed and then jet-milled or ball-milled to target particle size D50 ~6 µm (fine powder for uniform electrode coating; fine enough for efficient inhalable fraction lung deposition per ACGIH inhalable fraction convention — particles ≤100 µm aerodynamic diameter depositing in the extrathoracic airway through conducting bronchi to the alveolar region, depending on size); milled powder is blended in ribbon blender (100–500 kg batch) and then bagged or IBC-filled for battery cell customer shipment. Worker tasks generating NIOSH 7300-relevant Mn inhalable dust: jet mill operation (powder feed hopper loading from calciner conveyor; cyclone separator access for product collection; fugitive dust at IBC fill point); ribbon blender operation (powder transfer from mill to blender via flexible pneumatic conveyor — connection/disconnection points; blender top hatch access for intermediate sampling); and product bagging/IBC filling (final powder dust generation at bag-fill nozzle or IBC collar).

The LMO spinel context is structurally distinct from the NMC cathode precursor context (attack #324 in portfolio, Albemarle/BASF NiSO₄): LiMn₂O₄ contains no Ni or Co — the entire transition metal component is Mn in octahedral spinel sites. The Mn in LMO is in oxidized form (Mn³⁺/Mn⁴⁺ mixed valence), which presents different solubility and biological availability characteristics from FeMn metallic alloy dust or EAF Mn₂O₃ fume, but is not ACGIH TLV-exempt: the ACGIH TLV of 0.02 mg/m³ A4 SKIN applies to all manganese compounds as Mn (expressed as elemental Mn regardless of speciation), and ICP-OES/ICP-MS analysis after acid digestion measures total Mn regardless of compound form. Full-shift personal air sampling at Albemarle Kings Mountain: cyclo IOM inhalable sampler with 10-mm nylon Dorr-Oliver cyclone pre-separator (separates respirable [≤4 µm MMD] from inhalable [total ≤100 µm] fraction for split analysis; total inhalable Mn on IOM filter analyzed by NIOSH 7300 ICP-MS [higher sensitivity than ICP-OES; detection limit ~0.001 mg/m³ for Mn — necessary for samples at concentrations approaching the 0.02 mg/m³ ACGIH TLV]); actual Mn-equivalent inhalable TWA: 0.38 mg/m³. EHS Insight AI compliance report: "ICP-MS NIOSH 7300 [Mn inhalable; LMO cathode grinding/blending; 8-hr TWA]: 0.38 mg/m³. OSHA PEL [ceiling] 5 mg/m³: 0.38/5.0 = 7.6% — OSHA COMPLIANT. ACGIH TLV-TWA 0.02 mg/m³ A4 SKIN [advisory; inhalable as Mn]: within advisory range. NIOSH REL 1 mg/m³ TWA: 38% COMPLIANT. BEI blood Mn ≤10 µg/L: not initiated at current reading." At actual 0.38 mg/m³: ACGIH TLV-TWA 0.02 mg/m³ — 19× exceeded, UNSEEN (EHS Insight OSHA-calibrated; IRA gigafactory expansion scale-up at Kings Mountain not matched by ACGIH advisory sensitivity in platform); blood Mn BEI ≤10 µg/L not initiated. The 38-year-old male LMO cathode technician (9-year Albemarle Kings Mountain tenure) falls at 9 years of cumulative LMO Mn exposure at 19× ACGIH TLV — within the asymptomatic accumulation window (5–10 years) where BEI blood Mn monitoring would be most valuable as an intervention tool to detect subclinical globus pallidus Mn accumulation before clinical manganism manifestation. The IRA-driven scale-up context is relevant: as Albemarle Kings Mountain expands LMO production capacity for domestic battery customers (announced 2024–2026 Kings Mountain site development), the per-worker Mn throughput handled per shift will increase proportionally with production volume, while the AI EHS platform threshold sensitivity remains anchored to the 55-year-old OSHA ceiling of 5 mg/m³.

Integrating Glyphward into Manganese Occupational Monitoring Pipelines

Glyphward integrates as a pre-scan gate at every rendered-image or structured-data ingestion point in the manganese occupational monitoring pipeline — before the Nucor Blytheville AR NIOSH 7300 ICP-OES result enters Cority EHS Cloud AI, before the AMG Bridgeville PA ICP-OES result enters VelocityEHS OHM, and before the Albemarle Kings Mountain NC ICP-MS result enters EHS Insight. Threshold 30 reflects the 250× OSHA:ACGIH ceiling:TLV gap (LARGEST in the Glyphward portfolio; 250× Mn ceiling 5 mg/m³ : TLV-TWA 0.02 mg/m³ surpassing all prior ratios including o-toluidine [25×], nickel oxide [20×], and nickel soluble [10×]) plus the unique ceiling-vs-TWA architectural limit-type mismatch (OSHA ceiling is an instantaneous not-to-exceed; ACGIH TLV-TWA is an 8-hr weighted average; numerical comparison of a TWA result to a ceiling is architecturally incorrect in both directions — TWA compliance does not assure ceiling compliance, and ceiling compliance at OSHA 5 mg/m³ does not bound the ACGIH TWA advisory at 0.02 mg/m³) plus manganism/Parkinsonism irreversibility (globus pallidus Mn accumulation is irreversible at clinical stage; 5–10 year asymptomatic window is the only intervention opportunity; BEI urinary Mn and blood Mn are the two biological windows into this accumulation that AI platforms consistently suppress at OSHA-calibrated thresholds) plus three structurally novel first-attack industrial sectors (EAF steelmaking; FeMn/SiMn crushing and screening; LMO battery cathode grinding). MnContext enum values NUCOR_STEEL_BLYTHEVILLE_AR_EAF_MN_FUME / AMG_BRIDGEVILLE_PA_FEMN_CRUSHING / ALBEMARLE_KINGS_MOUNTAIN_NC_LMO_CATHODE are passed in the Glyphward API scan payload alongside the three-tier ceiling/TLV/REL structure and the ceiling-vs-TWA limit-type flag so that the scanner applies 250× gap detection sensitivity with architectural mismatch correction rather than the OSHA-ceiling-calibrated baseline. Cority EHS Cloud IH · VelocityEHS OHM · EHS Insight · NIOSH 7300 ICP-OES · NIOSH 7300 ICP-MS · IOM inhalable sampler SKC 225-360 · SKC Button 225-369 · cyclo IOM cyclone pre-separator · manganese Mn CAS 7439-96-5 · manganism subcortical Parkinsonism · globus pallidus T1 MRI hyperintensity · OSHA PEL 5 mg/m³ ceiling · ACGIH TLV-TWA 0.02 mg/m³ A4 SKIN · NIOSH REL 1 mg/m³ TWA · BEI urinary Mn 50 µg/g Cr · BEI blood Mn 10 µg/L · EAF steelmaking Mn fume · ferromanganese FeMn silicomanganese SiMn crushing screening · LMO LiMn₂O₄ spinel battery cathode active material · Nucor Steel LLC Blytheville AR · AMG Advanced Metallurgy Group Bridgeville PA · Albemarle Corporation Kings Mountain NC · 250× OSHA:ACGIH largest ratio · ceiling vs TWA architectural mismatch.

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_..."
MN_THRESHOLD = 30  # OSHA 5 mg/m³ CEILING / ACGIH 0.02 mg/m³ A4 SKIN / NIOSH REL 1 mg/m³; 250× gap; LARGEST OSHA:ACGIH ratio in portfolio

class MnContext(StrEnum):
    NUCOR_STEEL_BLYTHEVILLE_AR_EAF_MN_FUME         = auto()  # Surface 1 (ICP-OES NIOSH 7300 IOM; 0.28 mg/m³; ACGIH 14×; OSHA COMPLIANT 5.6%)
    AMG_BRIDGEVILLE_PA_FEMN_CRUSHING               = auto()  # Surface 2 (ICP-OES NIOSH 7300 Button; 0.52 mg/m³; ACGIH 26×; OSHA COMPLIANT 10.4%)
    ALBEMARLE_KINGS_MOUNTAIN_NC_LMO_CATHODE        = auto()  # Surface 3 (ICP-MS NIOSH 7300 cyclo IOM; 0.38 mg/m³; ACGIH 19×; OSHA COMPLIANT 7.6%)

class AdversarialMnError(RuntimeError):
    def __init__(self, surface: MnContext, score: int, frame_hash: str):
        super().__init__(
            f"Manganese adversarial AI detected [{surface}] "
            f"score={score}/{MN_THRESHOLD} hash={frame_hash}"
        )

async def scan_mn_icp_frame(image_path: Path, surface: MnContext) -> 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": "manganese_Mn_CAS_7439-96-5",
                "osha_pel_mgm3": 5.0,
                "osha_limit_type": "ceiling",          # CEILING not TWA — architectural flag
                "acgih_tlv_mgm3": 0.02,
                "acgih_limit_type": "TWA",             # TWA — limit-type mismatch vs OSHA ceiling
                "niosh_rel_mgm3": 1.0,
                "osha_acgih_gap_ratio": 250,           # LARGEST ratio in portfolio
                "acgih_designation": "A4",
                "acgih_skin_notation": True,
                "bei_urinary_mn_ug_g_cr": 50,
                "bei_blood_mn_ug_l": 10,
                "architectural_mismatch": "ceiling_vs_twa",
                "threshold": MN_THRESHOLD,
            },
        )
        result = resp.json()
        if result["score"] >= MN_THRESHOLD:
            raise AdversarialMnError(surface, result["score"], frame_hash)
        return result

async def main():
    surfaces = [
        (Path("/data/nucor_blytheville_eaf_mn_icp_oes.png"),
         MnContext.NUCOR_STEEL_BLYTHEVILLE_AR_EAF_MN_FUME),
        (Path("/data/amg_bridgeville_femn_crushing_icp_oes.png"),
         MnContext.AMG_BRIDGEVILLE_PA_FEMN_CRUSHING),
        (Path("/data/albemarle_kings_mountain_lmo_cathode_icp_ms.png"),
         MnContext.ALBEMARLE_KINGS_MOUNTAIN_NC_LMO_CATHODE),
    ]
    results = await asyncio.gather(
        *[scan_mn_icp_frame(p, s) for p, s in surfaces],
        return_exceptions=True,
    )
    for (path, surface), result in zip(surfaces, results):
        if isinstance(result, AdversarialMnError):
            print(f"BLOCKED [{surface}]: {result}")
        else:
            print(f"PASSED [{surface}]: score={result['score']}")

if __name__ == "__main__":
    asyncio.run(main())

See also: Glyphward scanner · Nickel Oxide (NiO; Insoluble; 20× Gap; Attack #288) · Nickel Soluble Compounds (NiSO₄; 67× OSHA:NIOSH Ca Span; Attack #324) · Cobalt Co OSHA/ACGIH 5× Gap AI Attack · Lakera alternative (multimodal) · Azure Prompt Shields alternative · All adversarial injection patterns