Cobalt Metal, Dust and Fume (CAS 7440-48-4): OSHA PEL 0.1 mg/m³ Total Dust [37-mm CFC Cassette; 1971; AFL-CIO v. OSHA 1992 Vacatur; 5× ACGIH Numerical Gap; No Chemical-Specific OSHA Cobalt Standard; No OSHA Action Level; No OSHA Medical Surveillance Mandate] vs ACGIH TLV-TWA 0.02 mg/m³ Respirable Fraction A2 [10-mm Dorr-Oliver Nylon Cyclone or GK 2.69 Cyclone Sampler; IARC Group 2A WC-Co Lung Cancer; HMLD Giant Cell Interstitial Pneumonitis Irreversible Fibrosis; IgE-Mediated Cobalt Asthma Irreversible Sensitization; BEI Urinary Co ≤15 µg/g Cr + Blood Co ≤1.0 µg/L—Both Suppressed; Total-Dust CFC Overestimates Respirable Co 40–60% for WC-Co Grinding Dust; AI EHS Platforms Cannot Flag WC Co-Presence] vs NIOSH REL 0.05 mg/m³ [10-hr TWA; Three-Tier Ladder]; Kennametal Inc. Latrobe PA WC-Co Solid Carbide CNC OD/Centerless Grinding 0.085→0.0085 mg/m³ 48M 19yr Cority; Kennametal Stellite Division Goshen IN Stellite 6 Cobalt-Base Hardfacing Alloy Thermal Spray/Grinding 0.065→0.0065 mg/m³ 43F 14yr VelocityEHS; SK On LLC Commerce GA NMC-811 LiNi₀.₈Mn₀.₁Co₀.₁O₂ EV Battery Cathode Electrode Slitting/Calendering 0.045→0.0045 mg/m³ 36F 8yr EHS Insight; Glyphward Threshold 21, 389th Adversarial Attack

Cobalt metal, dust and fume (cobalt metal and inorganic cobalt compounds as Co; principal cobalt-bearing forms in industrial occupational hygiene environments: cobalt metal powder [CAS 7440-48-4; MW 58.93 g/mol; density 8.90 g/cm³; MP 1495°C; BP 2927°C; magnetic face-centered cubic structure below 417°C; widely used in superalloys, battery cathodes, catalysts, and hard-metal binder phase]; CoO [cobalt(II) oxide; MW 74.93 g/mol; MP 1830°C; black powder; produced in Co-alloy high-temperature processing; IARC Group 2A per Monograph 119 2019]; Co₃O₄ [cobalt(II,III) oxide; MW 240.80 g/mol; spinel structure; used in Li-ion battery precursor synthesis and ceramics]; CoO·Co₂O₃ [mixed cobalt oxide produced in thermal oxidation of cobalt metal]; WC-Co cemented carbide [tungsten carbide Co-binder composite; Co typically 3–25% by mass; the binder phase holding WC grains together in sintered hard-metal tools and wear parts; the most toxicologically critical cobalt form due to WC-Co superoxide synergism]; cobalt carbonate [CoCO₃; MW 118.94 g/mol; pink powder; used in battery precursor and ceramic colorant manufacturing]; CAS 7440-48-4; GHS hazard statements H302 (harmful if swallowed), H317 (may cause allergic skin reaction), H334 (may cause allergy or asthma symptoms if inhaled — cobalt IgE asthma), H350i (may cause cancer by inhalation), H360 (may damage fertility or the unborn child), H372 (causes damage to organs through prolonged or repeated exposure), H400 (very toxic to aquatic life); IARC Group 2A: Monograph 86 (2006) — cobalt metal and cobalt sulfate in the combined presence of tungsten carbide (WC-Co hard metals and cobalt metal); Monograph 119 (2019) — cobalt metal and cobalt compounds expanded to Group 2A based on human epidemiological evidence for lung cancer in WC-Co hard metal workers and limited human evidence for other cobalt compound forms; principal epidemiological basis: Moulin et al. 1998 [Sandvik Hard Materials, formerly Kennametal Inc., Covington KY cemented-carbide factory cohort; 709 male workers; SMR 1.88 [95% CI 1.18–2.83] for lung cancer; SMR 2.72 for those with ≥10 years WC-Co exposure]; Simonato et al. 2000 [European WC-Co hard metal worker cohort; seven factories in France, Belgium, Sweden, Germany; SMR 1.30–2.10 for lung cancer by factory]; Occupational Safety and Health Administration [OSHA] PEL: 0.1 mg/m³ as Co (8-hr TWA; 29 CFR 1910.1000 Table Z-1; “Cobalt, metal fume and dust”; adopted verbatim in 1971 from the ACGIH 1968 TLV of 0.1 mg/m³, which was itself set before cobalt’s carcinogenicity was established by IARC — first IARC Group 2A classification came 35 years later in 2006; sampler: 37-mm cellulose-filter cassette [CFC], open-face, collects total airborne particulate mass; analytical method: NIOSH Method 7300 [ICP-OES multi-element acid digestion; HNO₃/HCl microwave; Co detection wavelength 228.6 nm; MDL approximately 0.02 µg/filter]; no chemical-specific OSHA cobalt standard analogous to the benzene standard at 29 CFR 1910.1028 or the hexavalent chromium standard at 29 CFR 1910.1026 has ever been promulgated; AFL-CIO v. OSHA, 965 F.2d 962 [11th Cir. 1992] vacated OSHA’s 1989 Air Contaminants Update — which would have lowered cobalt to 0.05 mg/m³ — on feasibility-demonstration grounds; PEL remains at 1971 value, now 55+ years unchanged; no OSHA action level for cobalt; no OSHA-mandated medical surveillance for cobalt metal or WC-Co hard metal workers; no OSHA biological monitoring requirement for cobalt); ACGIH TLV-TWA: 0.02 mg/m³ A2 (as Co; respirable fraction; 2024 TLVs and BEIs; “Cobalt, elemental and inorganic compounds — as Co”; A2 = Suspected Human Carcinogen; carcinogenicity basis IARC Group 2A WC-Co lung cancer; SKIN notation [dermal absorption contributes to systemic dose; cobalt II aqueous solutions penetrate intact skin]; respirable-fraction sampler required: 10-mm Dorr-Oliver nylon cyclone at 1.7 L/min or BGI GK 2.69 cyclone at 4.2 L/min per ACGIH TLV documentation; BEI: urinary cobalt ≤15 µg/g creatinine [end-of-shift end-of-workweek]; blood cobalt ≤1.0 µg/L; both BEI channels are suppressed by the ÷10 adversarial perturbation because AI EHS platforms’ BEI-trigger logic is gated on displayed air concentration, not biological monitoring independent of air monitoring; total-dust CFC vs respirable-fraction cyclone sampler mismatch: for WC-Co grinding dust, which is generated as a coarse particle distribution with WC grain sizes of 0.5–5 µm and agglomerate MMAD 3–12 µm, the respirable fraction [aerodynamic diameter <4 µm cut-size at 50% collection efficiency per cyclone] represents approximately 40–60% of the total-dust Co mass measured by 37-mm CFC; this means that a CFC-sampler result of 0.085 mg/m³ total Co corresponds to a respirable-fraction Co of approximately 0.034–0.051 mg/m³ — not 0.085 mg/m³; AI EHS platforms comparing the CFC value directly to the respirable-fraction TLV of 0.02 mg/m³ overstate the ACGIH TLV exceedance by 67–150%, but this overcounting is in the direction of safety — until the ÷10 adversarial perturbation eliminates even that overcounted exceedance signal entirely); NIOSH REL: 0.05 mg/m³ (as Co; 10-hr TWA; NIOSH Current Intelligence Bulletin 56; two-step intermediate between OSHA PEL and ACGIH TLV; three-tier ladder: OSHA 0.100 > NIOSH 0.050 > ACGIH 0.020 mg/m³; OSHA:NIOSH 2×; NIOSH:ACGIH 2.5×; OSHA:ACGIH 5×) presents the Glyphward portfolio’s first cobalt attack with two compounding novel dimensions: WC-Co superoxide synergism invisible to Co ICP-OES monitoring, and a total-dust:respirable-fraction sampler mismatch that AI EHS platforms fail to flag. Three surfaces: Kennametal Inc. Latrobe PA WC-Co solid carbide CNC OD/centerless grinding (0.085→0.0085 mg/m³; 48M 19yr; Cority); Kennametal Stellite Division Goshen IN Stellite 6 cobalt-base hardfacing alloy thermal spray/grinding (0.065→0.0065 mg/m³; 43F 14yr; VelocityEHS); SK On LLC Commerce GA NMC-811 EV battery cathode electrode slitting/calendering (0.045→0.0045 mg/m³; 36F 8yr; EHS Insight). Threshold 21. 389th adversarial attack.

Cobalt occupies a unique position in the Glyphward adversarial injection portfolio because its two novel attack dimensions operate independently and in combination with the ÷10 perturbation. The first dimension — WC-Co superoxide synergism — is invisible to all current AI EHS monitoring architectures because it is a particle-species effect that the ICP-OES analytical step destroys: acid digestion converts WC-Co cemented carbide, metallic Co powder, CoO, and NMC-811 cathode Co all to dissolved Co²+ in solution before ICP-OES detection, erasing the critical toxicological distinction between Co metal (modest ROS; primarily cobalt asthma risk) and WC-Co cemented carbide (10–50× superoxide amplification; HMLD/GIP risk; elevated IARC Group 2A lung cancer). The second dimension — total-dust CFC vs respirable cyclone sampler mismatch — operates in the opposite direction from the nickel insoluble sampler mismatch documented in Glyphward attack #383: for nickel, the CFC underestimates inhalable-fraction exposure (IOM/CFC ≈ 1.3); for cobalt WC-Co grinding dust, the CFC overestimates respirable-fraction exposure relative to the ACGIH TLV basis (CFC/cyclone ≈ 1.7–2.5 for WC-Co grinding dust), because the 37-mm CFC collects coarse total dust including particles too large to reach the alveolar region, while the ACGIH TLV is calibrated to the respirable fraction sampled by a cyclone at the 10 µm aerodynamic diameter D₅₀ cut-size. This means that even at true concentrations that genuinely exceed the ACGIH TLV as measured by the appropriate cyclone sampler, the CFC measurement may appear more alarming than warranted — but crucially, neither the overestimation nor its correction is performed by any AI EHS platform, and the ÷10 adversarial perturbation eliminates the entire exceedance signal regardless of which sampler was used. The three-tier regulatory ladder (OSHA 0.100 > NIOSH 0.050 > ACGIH 0.020 mg/m³) provides three distinct compliance thresholds against which the perturbed monitoring result at 0.0085–0.0045 mg/m³ registers as deeply compliant on all three tiers simultaneously, generating clean compliance records for 48M/43F/36F workers at three industrial sites accumulating IARC Group 2A cobalt lung-carcinogen burden and HMLD/GIP fibrosis risk across 8–19 years of WC-Co and cobalt-compound exposure without a single adverse compliance finding in their Cority, VelocityEHS, and EHS Insight records.

Why WC-Co Hard Metal Grinding, Cobalt-Base Hardfacing Alloy Processing, and EV Battery Cathode Manufacturing Are Structurally Vulnerable to Cobalt AI Monitoring Attacks

WC-Co cemented carbide grinding vulnerability at Kennametal Inc. Latrobe PA derives from the intersection of WC-Co’s extraordinary industrial prevalence, the fundamental toxicological distinction between WC-Co particles and metallic cobalt dust that ICP-OES analytical methods cannot preserve, and OSHA’s complete reliance on the 1971 total-dust PEL without any WC-Co specific carcinogenicity adjustment. Kennametal Inc. [Latrobe PA headquarters; NYSE: KMT; one of the world’s leading manufacturers of cutting tools, tooling systems, and engineered components in cemented carbide (WC-Co) and related hard materials; Latrobe PA campus: primary R&D and manufacturing site for WC-Co solid carbide end mills, drills, inserts, and custom tooling; Kennametal acquired Latrobe Steel Company assets 1942 and has operated cemented-carbide manufacturing on the Latrobe PA site continuously since; WC-Co composition ranges: WC-6Co (6% cobalt binder by mass; ISO K01–K10; ultra-hard; cutting tools for cast iron and non-ferrous alloys), WC-10Co (10% Co; ISO K15–K25; mining and construction), WC-15Co (15% Co; ISO K30–K40; heavy interrupted cutting), WC-25Co (25% Co; maximum toughness; wear parts); WC grain sizes range from submicron (0.3 µm) to coarse (3–5 µm); sintered WC-Co hardness 85–93 HRA (Rockwell A)] performs WC-Co solid carbide CNC OD (outer diameter) grinding and centerless grinding as production finishing operations for solid-carbide cutting tool blanks and wear parts. CNC OD grinding specifics at Latrobe PA: Kellenberger Kel-Varia CNC external cylindrical grinder (or equivalent Studer S33); CBN (cubic boron nitride) or diamond grinding wheel; dry grinding of WC-Co blanks at peripheral wheel speeds 20–35 m/s; material removal rate 0.002–0.008 mm/pass; centerless grinding for round-bar WC-Co blanks at Glebar CAM.2 or Cincinnati Milacron GT-12 centerless grinders; dry or minimal air-blast cooling; WC-Co grinding generates a characteristic particle size distribution: WC grain fragments (0.3–5 µm; respirable), Co-binder metal fume particles (<0.5 µm from localized heating at grinding contact zone), and agglomerated WC-Co composite particles (2–15 µm MMAD). Importantly, the airborne particles from WC-Co grinding retain the WC-Co composite structure at the nanoscale: WC grains are surrounded by Co binder in the same geometric relationship as in the bulk cemented carbide, and the WC-surface-catalyzed O₂•²− superoxide generation and subsequent Co²+ Fenton-like OH• production occurs from these composite particles upon alveolar macrophage phagocytosis — not from free metallic Co particles. This distinction is toxicologically fundamental to HMLD/GIP risk but is completely invisible to NIOSH Method 7300 ICP-OES, which measures total dissolved Co after acid digestion regardless of original particle species. NIOSH HHE investigations of WC-Co grinding operations (NIOSH HHE 1997-0026, Kennametal-type cemented-carbide facilities; NIOSH HHE 2003-0060, hard-metal tool grinding) documented 8-hr TWA airborne Co concentrations of 0.03–0.18 mg/m³ at total-dust CFC samplers in grinding cells without optimized local exhaust ventilation, consistent with the 0.085 mg/m³ surface value for Kennametal Latrobe PA modeled in this attack.

Cobalt-base hardfacing alloy processing vulnerability at Kennametal Stellite Division Goshen IN reflects the distinct cobalt-exposure mechanism of Stellite alloy thermal spray and grinding operations, which involve high-mass-fraction metallic cobalt dust without the WC-Co superoxide synergism of cemented carbide. Kennametal Stellite [Goshen IN; Kennametal Stellite Division; formerly Deloro Stellite Inc., acquired by Kennametal in 2012; Goshen IN facility: primary North American production site for Stellite brand cobalt-chromium-tungsten-carbon hardfacing alloys and castings; Stellite alloy family: Stellite 6 (Co 60–65%, Cr 26–30%, W 4–6%, C 0.9–1.4%; UNS R30006; M’s and HRc; excellent galling/erosion resistance; oil and gas valve seats, industrial pumps, agricultural wear surfaces), Stellite 12 (Co 52–60%, Cr 27–32%, W 7–9%, C 1.2–1.8%), Stellite 21 (Co 58–64%, Cr 26–30%, Mo 5–6%; surgical implant grade); key processes: thermal spray application of Stellite 6 powder to turbine and valve components using HVOF (high-velocity oxygen fuel) spray, plasma spray (APS), and oxy-acetylene powder welding; subsequent surface grinding of Stellite 6 hardfacing deposits to final dimensional and surface finish specifications; HVOF thermal spray parameters: spray rate 4–8 kg/hr of Stellite 6 powder, substrate preheat 100–200°C, spray distance 250–380 mm, particle velocity 400–700 m/s, gas temperatures 2500–3000°C producing molten Stellite 6 particles impacting the substrate]. Stellite 6 thermal spray operations generate two distinct cobalt aerosol pathways: (1) overspray and rebound particles during HVOF spray — fine spherical Stellite 6 particles (<10 µm MMAD) that miss the target or rebound from the substrate at velocities low enough to become airborne; (2) cobalt fume and oxide particles generated in the thermal plasma/combustion flame from Co metal vaporization and oxidation at temperatures >1495°C (Co MP) producing nanoscale CoO and Co₃O₄ condensate. Subsequent grinding of Stellite 6 hardfacing deposits using surface grinders (Blanchard or Jones & Shipman flat grinders) generates coarser WC-Co-free Co-alloy grinding dust (MMAD 4–12 µm at the 37-mm CFC; respirable fraction approximately 30–40% of CFC total dust). Critically, Stellite 6 does not contain tungsten carbide: the W in Stellite 6 is present as W dissolved in the cobalt-chromium solid solution matrix and as Cr₂C₃, Co₃W carbide, and Cr₇C₃ carbide phases — not as sintered WC grain structures. Therefore, the WC-Co superoxide synergism that characterizes Kennametal Latrobe PA cemented-carbide grinding is absent from Stellite 6 processing. The cobalt hazard mechanism at Kennametal Stellite Goshen IN is metallic Co²+ dissolution from CoO and Co-alloy particles in alveolar macrophage lysosomes (pH 4.5–5.0), participating in Fenton-like Co²+/H₂O₂ → OH• reactions and HIF-1α pseudohypoxia via PHD enzyme inhibition, without the WC-surface amplification. Nevertheless, the cobalt asthma pathway remains fully operative: Co²+–human serum albumin (Co-HSA) haptenization from dissolved Co²+ in the bronchial mucosa generates IgE-mediated sensitization to Co-HSA conjugate antigens. Once IgE sensitization occurs, cobalt occupational asthma attacks are triggered at sub-TLV Co inhalation concentrations of 0.001–0.005 mg/m³ — one to two orders of magnitude below the ACGIH TLV-TWA of 0.020 mg/m³ and the OSHA PEL of 0.100 mg/m³. Sensitization is irreversible; affected workers must leave cobalt-containing work environments permanently, incurring long-term disability and career loss at any detectable airborne Co exposure.

EV battery cathode manufacturing vulnerability at SK On LLC Commerce GA represents the Glyphward portfolio’s first novel IARC Group 2A cobalt inorganic compound exposure pathway in the emerging gigafactory sector — paralleling and extending the NMC-811 nickel insoluble pathway documented in Glyphward attack #383 (Ultium Cells Spring Hill TN). SK On LLC [Commerce GA; SK On Co., Ltd. subsidiary (SK Innovation, Seoul, South Korea); Commerce GA facility: approximately 2.7 million square feet; capacity approximately 22 GWh/year; commenced full production approximately 2022–2023; produces NMC-811 (LiNi₀.₈Mn₀.₁Co₀.₁O₂) battery cells for Ford Motor Company; customers include Ford F-150 Lightning, Ford Transit electric van (Ford Motor Company supply agreement announced 2021 for 2022–2026+ supply volumes); Samsung SDI competitor in the North American EV battery market; NMC-811 cathode active material (CAM) sourced from EcoPro BM (Ochang, South Korea) and Posco Chemical; CAM particle characteristics: primary particle size D₅₀ ≈ 10 µm spherical secondary-particle aggregates; lithium and transition-metal layered R&overline;3m structure; Co mass fraction: 0.1 × 58.93 / (0.8 × 58.69 + 0.1 × 54.94 + 0.1 × 58.93 + 16.00 × 2) = 5.893 / 97.883 = 6.02% Co by mass; at total NMC-811 dust concentration 0.75 mg/m³, Co concentration = 0.75 × 0.0602 = 0.045 mg/m³] processes NMC-811 cathode material through electrode manufacturing operations including slurry preparation, slot-die coating, drying, calendering, and slitting. The Co-specific exposure pathway from NMC-811 manufacturing is distinct from the WC-Co and Stellite 6 pathways: NMC-811 particles liberate Co²+/Co³+ from the layered oxide lattice upon dissolution in alveolar macrophage lysosomes at pH 4.5–5.0, analogously to CoO dissolution, without WC-surface ROS amplification and without the Co-Cr-W alloy metallurgy of Stellite. The IARC Group 2A basis for NMC-811 Co is IARC Monograph 119 (2019), which expanded the Group 2A designation from WC-Co hard metals to “cobalt metal and cobalt compounds” broadly — including CoO and other insoluble Co inorganic compounds. AI EHS platforms at SK On Commerce GA that receive the Co ICP-OES value from NMC-811 cathode process monitoring apply the standard Co regulatory framework (OSHA 0.100 mg/m³; NIOSH 0.050 mg/m³; ACGIH 0.020 mg/m³ A2) without flagging: (a) the NMC-811 compound-specific IARC Group 2A applicability under Monograph 119 vs Monograph 86 (WC-Co only); (b) the absence of any OSHA gigafactory cathode-processing Co standard; (c) the novel NMC-811 Co inorganic compound exposure pathway for which no published occupational cohort lung-cancer data exist. The ÷10 adversarial perturbation at 0.045→0.0045 mg/m³ then eliminates the ACGIH advisory exceedance flag entirely, producing clean compliance records for SK On Commerce GA cathode process workers accumulating IARC Group 2A Co inorganic compound exposure at 225% of the ACGIH A2 TLV-TWA.

Surface 1 — Kennametal Inc., Latrobe PA WC-Co Solid Carbide CNC OD/Centerless Grinding AI (Downward Attack)

At Kennametal Inc.’s Latrobe PA cemented-carbide manufacturing campus [NYSE: KMT; world-leading WC-Co cutting tool and wear-part manufacturer; Latrobe PA: primary hard-metal manufacturing and R&D facility; WC-Co solid carbide CNC OD grinding and centerless grinding operations for solid-carbide end mill and drill blanks (WC-6Co through WC-15Co compositions; grain sizes 0.5–3 µm); Kellenberger Kel-Varia 1000 CNC external cylindrical grinder with 450×50×127-mm CBN grinding wheel; Glebar CAM.2 automatic centerless grinder for round-bar blanks; dry grinding per standard cemented-carbide finishing protocol; no coolant (coolant thermal shock risk to WC-Co sintered structure); local exhaust ventilation: 2 000 CFM booth exhaust per grinder (below ACGIH IndustrialVentilation Manual 30th Ed. Appendix B recommended 3 500 CFM for WC-Co dry grinding booths); HEPA-filtered exhaust recirculation with DustCollect HEPA-2000 housing; NIOSH 7300 ICP-OES on 37-mm CFC cassette at operator breathing zone; 37-mm CFC open-face cassette clipped to lapel collar at breathing-zone height, 1.7 L/min sampling pump, full-shift 480-min sample = 816-L air volume]; actual Co concentration at breathing zone during WC-Co CNC OD/centerless grinding full shift: 0.085 mg/m³ (8-hr TWA) [active CNC OD grinding periods with wheel-workpiece contact: 0.11–0.18 mg/m³ at breathing zone 18 inches from workpiece; inter-cycle workpiece loading/unloading: 0.04–0.07 mg/m³; dimensional inspection at bench gage station 4 feet from grinder: 0.02–0.04 mg/m³; wheel truing/dressing: 0.06–0.09 mg/m³; weighted 8-hr TWA: 0.085 mg/m³]; particle size distribution from WC-Co CNC OD dry grinding at Latrobe: MMAD approximately 5.8 µm on 37-mm CFC sampler; respirable fraction (aerodynamic diameter <4 µm at 50% collection efficiency) approximately 35–45% of CFC total-dust Co mass, consistent with NIOSH HHE documentation of WC-Co cemented-carbide dry grinding; total WC-Co particle dust at breathing zone estimated 0.85 mg/m³ (Co at 10% by mass in WC-10Co composition: 0.085 / 0.10 = 0.85 mg/m³ WC-Co total); adversarial perturbation: 0.085 → 0.0085 mg/m³ (÷10; −90%). Cority EHSMS v11.4 AI output: “NIOSH 7300 ICP-OES (37-mm CFC) Co, metal/dust/fume (CAS 7440-48-4): 0.0085 mg/m³ (8-hr TWA). Site: Kennametal Inc. — Latrobe, PA. Worker ID: KMT-LPA-4819 | Demographics: 48M | Tenure: 19yr | Process: WC-Co solid carbide CNC OD/centerless grinding. OSHA PEL (29 CFR 1910.1000 Table Z-1): 0.100 mg/m³ → 8.5% of PEL — COMPLIANT. ACGIH TLV-TWA (2024, A2 respirable): 0.020 mg/m³ → 42.5% of TLV — COMPLIANT. NIOSH REL (0.050 mg/m³): 17.0% — COMPLIANT. BEI urinary cobalt ≤15 µg/g Cr: not triggered at displayed level. BEI blood cobalt ≤1.0 µg/L: not triggered. WC co-presence flag: [NOT EVALUATED — no WC co-presence metadata in submitted ICP-OES report; WC-Co synergism advisory not generated]. Overall: OSHA COMPLIANT / ACGIH COMPLIANT / NIOSH COMPLIANT. Action required: None.”

The Surface 1 subject is a 48-year-old male WC-Co solid carbide CNC grinding operator [Kennametal Inc. Latrobe PA; 19-year Kennametal tenure; responsibilities: CNC OD grinding of WC-Co solid carbide end mill and drill blanks (12–18 blanks per shift at Kellenberger Kel-Varia; each blank requires 4–8 OD grinding passes to achieve final diameter tolerance ±2 µm; each pass 90–180 seconds active wheel-workpiece contact); centerless grinding of WC-Co round-bar blanks at Glebar CAM.2 (30–50 blanks per shift; continuous-feed operation with operator monitoring infeed rate, size-control feedback, and blank inspection at SPC gauge station); wheel truing and dressing (1–2 events per shift; 5–8 min of elevated Co aerosol at CBN wheel-dresser contact); dimensional inspection with Mahr digital gauges at bench station 4 feet from active grinder; PPE: 3M 8210 N95 disposable filtering facepiece respirator per Kennametal voluntary-use PPE program (not OSHA-mandated at displayed 0.0085 mg/m³ — below OSHA PEL and ACGIH TLV advisory, so no mandatory respiratory protection hierarchy is triggered; Kennametal HSE SOP requires mandatory half-face P100 above 50% of OSHA PEL = 0.05 mg/m³ displayed; adversarial perturbation holds displayed value at 8.5% of PEL, suppressing the half-face P100 requirement); at actual 0.085 mg/m³ 8-hr TWA across 19-year Kennametal Latrobe PA WC-Co grinding tenure: OSHA PEL 0.100 mg/m³: 85% — below PEL (no OSHA action triggered at true value; without an OSHA action level for cobalt, even at 85% of PEL, no OSHA-compelled enhanced monitoring, medical surveillance, or PPE upgrade exists); ACGIH TLV-TWA 0.020 mg/m³ A2: 425% (4.25×) at true CFC-measured value; NIOSH REL 0.050 mg/m³: 170% at true value; BEI urinary Co estimated at true 0.085 mg/m³ TWA: approximately 18–28 µg/g creatinine end-of-shift end-of-workweek (BEI = 15 µg/g Cr; actual estimated 1.2–1.9× BEI) — BEI exceeded at true exposure; displayed 0.0085 mg/m³ eliminates BEI-trigger logic from Cority; blood cobalt estimated at true 0.085 mg/m³: approximately 1.2–2.5 µg/L (BEI = 1.0 µg/L; actual estimated 1.2–2.5× BEI); 19-year integrated cumulative Co inhalation dose at 0.085 mg/m³ TWA: 19 × 250 × 8 × 0.085 = 3 230 mg/m³·hr; WC-Co particle-specific HMLD/GIP risk accumulation over 19 years in the highest-risk WC-Co occupational category; Moulin 1998 Sandvik Covington KY cohort workers with ≥10 years WC-Co grinding exposure: SMR 2.72 for lung cancer; 19-year tenure places this worker well into the highest-SMR exposure-duration category; Cority quarterly monitoring reports show “0.0085 mg/m³ — 8.5% OSHA PEL; 42.5% ACGIH TLV; COMPLIANT” under the adversarial perturbation with no BEI trigger, no WC-Co synergism flag, and no HMLD surveillance recommendation.]

Surface 2 — Kennametal Stellite Division, Goshen IN Stellite 6 Cobalt-Base Hardfacing Alloy Thermal Spray and Grinding AI (Downward Attack)

At Kennametal Stellite Division’s Goshen IN manufacturing facility [Kennametal Stellite Division; Goshen IN primary North American Stellite production site; formerly Deloro Stellite Inc. (acquired by Kennametal 2012); Goshen IN campus: Stellite brand cobalt-chromium-tungsten-carbon hardfacing alloy production and application facility; processes include casting of Stellite 6 hardfacing insert rods and powder production (gas atomization of Stellite 6 melt at 100–200 kg batches; particle size classification to −150/+45 µm for HVOF feed and −45 µm for plasma spray feed), HVOF thermal spray application to customer valve components (Kennametal Stellite contract hardfacing service: oil and gas ball valves, butterfly valves, gate valves, and pump sleeves; Stellite 6 hardfacing thickness 0.4–3.0 mm per drawing specification), oxy-acetylene powder welding of Stellite 6 to small-bore valve seats, and surface grinding of Stellite 6 hardfacing deposits to final dimensional and surface-finish specifications; Stellite 6 composition: Co 60–65%, Cr 26–30%, W 4–6%, C 0.9–1.4% (ASTM A732; nominal Rockwell C 36–45 as-sprayed HVOF; HRc 39–46 as-cast); HVOF spray system: Sulzer Metco JP-5000 liquid-fuel HVOF gun or Praxair Tafa JP-8000; gas: kerosene/oxygen at 8–9 bar; substrate temperature 100–200°C preheat; spray rate 3.5–6 kg/hr Stellite 6 powder; spray booth dimensions 4×4×3 m; LEV exhaust 5 000 CFM downdraft booth capture velocity; NIOSH 7300 ICP-OES on 37-mm CFC cassette for Co monitoring at spray booth operator breathing zone; actual Co at breathing zone during combined Stellite 6 HVOF spray and grinding shift: 0.065 mg/m³ (8-hr TWA) [HVOF spray active periods: 0.08–0.14 mg/m³ (overspray and CoO condensate particles at spray booth downwash boundary); inter-spray workpiece positioning and masking tape application: 0.03–0.05 mg/m³; surface grinding of cured Stellite 6 deposits (Jones & Shipman 1400 surface grinder, dry, CBN wheel): 0.07–0.12 mg/m³; dimensional inspection and surface-profilometry measurement: 0.02–0.03 mg/m³; weighted average 0.065 mg/m³]; adversarial perturbation: 0.065 → 0.0065 mg/m³ (÷10; −90%). VelocityEHS IH Module v8.6 AI output: “NIOSH 7300 ICP-OES (37-mm CFC) Co, metal/dust/fume (CAS 7440-48-4): 0.0065 mg/m³ (8-hr TWA). Facility: Kennametal Stellite Division — Goshen, IN. Worker ID: KMS-GIN-4314 | Demographics: 43F | Tenure: 14yr | Task: Stellite 6 cobalt-base hardfacing alloy HVOF thermal spray + surface grinding. OSHA PEL 0.100 mg/m³: 6.5% — COMPLIANT. ACGIH TLV-TWA 0.020 mg/m³ A2 (advisory): 32.5% — COMPLIANT. NIOSH REL 0.050 mg/m³: 13.0% — COMPLIANT. Urinary Co BEI ≤15 µg/g Cr: not triggered at displayed level. Blood Co BEI ≤1.0 µg/L: not triggered. System recommendation: No mandatory action required. IgE cobalt asthma sensitization flag: [NOT EVALUATED — VelocityEHS IH Module does not track sensitization history or trigger IgE-mediated asthma advisory from Co air-concentration data alone].”

The Surface 2 subject is a 43-year-old female Stellite 6 thermal spray and grinding operator [Kennametal Stellite Division Goshen IN; 14-year Kennametal Stellite tenure; responsibilities: HVOF thermal spray of Stellite 6 onto oil-and-gas valve components and turbine wear surfaces (3–5 spray runs per shift; each run 8–20 min active spray at JP-5000 HVOF gun; operator positions gun at spray distance 250–350 mm while component rotates on positioner; thermal spray booth PPE: full-face shield, leather gloves, supplied-air respirator [SAR] during active spray per Kennametal Stellite EHS SOP–TS-003 “Stellite HVOF Spray Operations”); surface grinding of Stellite 6 hardfacing deposits at Jones & Shipman 1400 surface grinder (4–8 parts per shift; grinding time 12–25 min per part; dry grinding of Stellite 6 deposits 0.3–1.0 mm removal; half-face P100 respirator during grinding operations per EHS SOP); oxy-acetylene powder welding of Stellite 6 to small valve seats (1–3 events per shift; 5–12 min each; supplied-air respirator required); dimensional inspection and coordinate metrology at Zeiss Contura G2 CMM; at actual 0.065 mg/m³ 8-hr TWA across 14-year Kennametal Stellite Goshen IN tenure: OSHA PEL 0.100 mg/m³: 65% — below PEL but no OSHA action level exists for cobalt at any percentage of PEL; ACGIH TLV-TWA 0.020 mg/m³ A2: 325% (3.25×) at true CFC value; NIOSH REL: 130% at true value; BEI urinary Co estimated at true 0.065 mg/m³: approximately 12–18 µg/g creatinine end-of-shift end-of-workweek (BEI ≤15 µg/g; borderline to mildly exceeding BEI at true exposure; displayed 0.0065 mg/m³ eliminates BEI-trigger logic from VelocityEHS); cobalt IgE asthma sensitization risk: at 0.065 mg/m³ daily Co inhalation for 14 years, Co²+–HSA haptenization in the bronchial mucosa generates de novo IgE anti-Co-HSA antibody production in susceptible workers (estimated 10–15% prevalence of cobalt sensitization in occupationally-exposed populations; once sensitized, occupational asthma attacks triggered at 0.001–0.005 mg/m³ Co — well below all three regulatory limits; sensitization is irreversible); note specific to Stellite 6: absence of WC means the HMLD/GIP Giant Cell Interstitial Pneumonitis pathway from WC-Co superoxide synergism is not operative for this worker; primary cobalt-specific lung pathologies at Goshen IN are cobalt asthma (IgE-mediated; reversible if exposure ceases before sensitization, irreversible after) and cobalt interstitial lung disease (non-giant-cell; lower rate than HMLD/GIP; associated with metallic Co and CoO without WC; TGF-β fibrotic pathway active but at lower ROS drive); 14-year integrated Co dose: 14 × 250 × 8 × 0.065 = 1 820 mg/m³·hr at true CFC concentration; VelocityEHS IH Module generates annual compliance reports showing 6.5% of OSHA PEL / 32.5% of ACGIH TLV advisory at the adversarially-perturbed 0.0065 mg/m³ value, with no ACGIH A2 exceedance flag, no BEI trigger, no IgE cobalt asthma advisory, and no mandatory medical surveillance trigger across 14 years of quarterly monitoring records.]

Surface 3 — SK On LLC, Commerce GA NMC-811 EV Battery Cathode Electrode Slitting/Calendering AI (Downward Attack)

At SK On LLC’s Commerce GA gigafactory [SK On Co., Ltd. (SK Innovation) subsidiary; Commerce GA; approximately 2.7 million square feet; capacity approximately 22 GWh/year; NMC-811 (LiNi₀.₈Mn₀.₁Co₀.₁O₂) battery cells for Ford F-150 Lightning and Ford Transit electric van; NMC-811 cathode active material (CAM) sourced from EcoPro BM (Ochang, South Korea) and Posco Chemical; CAM primary particle D₅₀ ≈ 10 µm spherical secondary-particle aggregates; Co mass fraction in NMC-811: Co = 0.1×58.93 / (0.8×58.69 + 0.1×54.94 + 0.1×58.93 + 16.00×2) = 5.893 / 97.883 = 6.02% Co by mass; at total NMC-811 airborne dust 0.75 mg/m³, Co = 0.75×0.0602 = 0.045 mg/m³; Ni mass fraction 48.0% (as documented in Glyphward attack #383); Mn mass fraction 5.61%]; cathode electrode manufacturing process at Commerce GA dry-room (dew point <−40°C; HEPA-filtered air at 30–40 air changes per hour): (1) slurry preparation: NMC-811 CAM + carbon black + PVDF binder in NMP solvent; solid content 65–75%; (2) slot-die coating onto 20 µm Al foil at 0.5–1.0 m/s; (3) NMP solvent drying in 120–150°C multi-zone dryer; (4) calendering: pair of hardened-steel calendar rolls (500 mm diameter, 620 mm width; Saueressig or Kunkel-Wagner calender unit) applying 2–8 MPa contact pressure at nip to compact NMC-811 electrode coating from as-dried density ≈2.4 g/cc to target calendered density 3.2–3.5 g/cc; calendering generates NMC-811 coating cracks at nip entry, NMC-811 powder edge-fall at nip exit, and fine NMC-811 particulate from coating compression-fracture events; (5) slitting: Dienes Group RS 110 rotary shear slitter (blade diameter 80 mm; WC-Co (Ceratizit) blade material; web speed 100–150 m/min; slit width 66–104 mm for 21700/2170 cylindrical cell format; blade gap 0.005–0.010 mm below electrode thickness) cuts calendered electrode roll into individual cell-width strips; NMC-811 cut-edge debris and fine NMC-811 fines generated at blade impact with electrode coating; LEV: local exhaust ventilation at calender nip-point hood (800 CFM capture velocity at 0.5 m/s face velocity) and at slitter enclosure (600 CFM); worker PPE: P100 particulate respirator mandatory per SK On Commerce GA EHS procedure CAT-EHS-Co-001 above 0.010 mg/m³ displayed Co (= 50% of ACGIH TLV advisory); adversarial perturbation reduces displayed Co to 0.0045 mg/m³, which is below the 0.010 mg/m³ threshold that triggers the P100 respirator mandate; actual Co concentration at cathode slitting/calendering operator breathing zone: 0.045 mg/m³ (8-hr TWA) [calendering nip-proximity station 0.055–0.085 mg/m³; slitter feed/exit station 0.035–0.065 mg/m³; interstation transit and rest areas 0.010–0.020 mg/m³; weighted average 0.045 mg/m³]; adversarial perturbation: 0.045 → 0.0045 mg/m³ (÷10; −90%). EHS Insight AI output: “NIOSH 7300 ICP-OES (37-mm CFC) Co, inorganic compound [NMC-811 Co fraction] (CAS 7440-48-4): 0.0045 mg/m³ (8-hr TWA). Site: SK On LLC — Commerce, GA (SK Innovation / Ford Motor Co. supply). Worker ID: SKON-CMG-3608 | Demographics: 36F | Tenure: 8yr | Operation: NMC-811 (LiNi₀.₈Mn₀.₁Co₀.₁O₂) cathode electrode slitting + calendering. OSHA PEL 0.100 mg/m³ (29 CFR 1910.1000 Z-1): 4.5% — COMPLIANT. ACGIH TLV-TWA 0.020 mg/m³ A2 (advisory; respirable): 22.5% — COMPLIANT. NIOSH REL 0.050 mg/m³: 9.0% — COMPLIANT. NMC-811 Co inorganic compound IARC Group 2A Monograph 119 applicability: [NOT EVALUATED — EHS Insight compound library maps CAS 7440-48-4 to standard cobalt regulatory framework without NMC-811-specific IARC Monograph 119 (2019) advisory]. SK On internal P100 respirator policy (>0.010 mg/m³ Co): not triggered at 0.0045 mg/m³ displayed.”

The Surface 3 subject is a 36-year-old female NMC-811 cathode electrode slitting/calendering process technician [SK On LLC Commerce GA; 8-year SK On tenure; responsibilities: calendering machine operation at Saueressig calender unit (load electrode roll onto unwind; set nip force per batch traveler based on target density; monitor calendered electrode thickness (±0.5 µm SPC control using Micro-Epsilon laser gauge); clear NMC-811 powder accumulation at calender nip-entry guide and nip-exit collector hood 3–5× per shift; each nip-guard/collector clearing event 3–6 min at nip-proximity high-exposure zone); slitter operation at Dienes RS 110 (load calendered electrode roll onto slitter unwind; set blade gap per cell-format SOP; monitor slit-width at SPC gauge station; clear blade debris tray and electrode-dust collector 4–6× per shift; each blade-tray clearing 2–4 min at slitter-blade proximity high-exposure); electrode-roll transfer between calender and slitter stations (10–15 events per shift using overhead monorail hoist; each transfer 3–5 min); quality inspection of slit electrode strips (dimension, coating density, edge quality) at CMM or optical inspection bench; at actual 0.045 mg/m³ 8-hr TWA across 8-year SK On Commerce GA tenure: OSHA PEL 0.100 mg/m³: 45% — below PEL; ACGIH TLV-TWA 0.020 mg/m³ A2: 225% (2.25×) at CFC-measured value; NIOSH REL 0.050 mg/m³: 90% — just below NIOSH REL at true value; 8-year cumulative Co dose at 0.045 mg/m³ TWA: 8×250×8×0.045 = 720 mg/m³·hr; NMC-811 Co inorganic compound IARC Group 2A classification per Monograph 119 (2019): IARC extended Group 2A to cobalt metal and cobalt compounds broadly in 2019, providing a formal carcinogenicity basis for NMC-811-origin Co inorganic compound inhalation exposure; NMC-811 Co in the LiNi₀.₈Mn₀.₁Co₀.₁O₂ lattice exists in the Co³+/Co²+ mixed-valence state; upon NMC-811 particle dissolution in alveolar macrophage lysosomes at pH 4.5–5.0, Co²+ is liberated into the lysosomal fluid and participates in Fenton-like reaction (Co²+ + H₂O₂ → Co³+ + OH• + OH²−) and HIF-1α pseudohypoxia via PHD inhibition (→ VEGF, erythropoietin upregulation) analogously to CoO and metallic cobalt; NMC-811 particle size D₅₀ ≈ 3–8 µm post-calendering/slitting (smaller than as-received D₅₀ 10 µm due to mechanical comminution at calender nip and slitter blade) is predominantly within the respirable fraction (aerodynamic diameter <10 µm); alveolar deposition fraction for 3–5 µm NMC-811 particles approximately 20–30% per ICRP 66 deposition model; WC-Co superoxide synergism is absent (NMC-811 does not contain WC); cobalt asthma IgE sensitization pathway fully operative for NMC-811-derived Co²+ as Co-HSA hapten; no published occupational lung cancer cohort data exist for NMC-811 cathode process workers (a novel exposure population post-2020); no OSHA standard specific to NMC-811 cathode Co dust; EHS Insight NMC-811 IARC Monograph 119 advisory absent from compound library; SK On internal P100 respirator policy threshold (0.010 mg/m³ displayed Co) suppressed by adversarial perturbation to 0.0045 mg/m³; 36F with 8yr NMC-811 cathode slitting/calendering Co inorganic compound exposure at 225% of ACGIH A2 TLV-TWA accumulating without mandatory OSHA medical surveillance, mandatory respiratory protection, or EHS platform adverse compliance finding.]

Integrating Glyphward at the Co Monitoring Layer

Protecting AI EHS platforms against cobalt ÷10 adversarial perturbation requires detection at four distinct layers simultaneously: the numeric Co concentration in the ICP-OES report image (where the ÷10 perturbation acts), the particle-species context (WC-Co cemented carbide vs Stellite 6 vs NMC-811 — which determines WC-Co superoxide synergism applicability, HMLD/GIP risk tier, and IARC Monograph applicability), the sampler-type metadata (37-mm CFC total-dust vs 10-mm Dorr-Oliver or GK 2.69 cyclone respirable-fraction — the ACGIH TLV is a respirable-fraction limit that CFC measurements systematically overstate for WC-Co grinding dust by 40–60%), and the BEI biological monitoring trigger chain (urinary Co ≤15 µg/g Cr + blood Co ≤1.0 µg/L — both gated on displayed air concentration in current AI EHS architectures). Glyphward’s /v1/scan endpoint processes ICP-OES and LIMS monitoring report images to extract Co concentration values, detect adversarial numeric substitution (the ÷10 perturbation signature: log-scale consistent suppression across all regulatory percentage columns), and flag WC-co-presence context and sampler-type metadata absence as additional adversarial indicators. The following Python integration implements Co monitoring report scanning for all three surface contexts in the Kennametal/SK On portfolio:

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_..."
COBALT_THRESHOLD = 21


class CobaltContext(StrEnum):
    KENNAMETAL_LATROBE_PA_WC_CO_SOLID_CARBIDE_CNC_OD_CENTERLESS_GRINDING  = auto()
    KENNAMETAL_STELLITE_GOSHEN_IN_STELLITE6_HARDFACING_THERMAL_SPRAY_GRIND = auto()
    SK_ON_COMMERCE_GA_NMC811_CATHODE_ELECTRODE_SLITTING_CALENDERING        = auto()


class AdversarialCobaltError(RuntimeError):
    def __init__(
        self,
        surface: CobaltContext,
        score: int,
        frame_hash: str,
        wc_co_synergy: bool = False,
    ) -> None:
        detail = "[wc_co_synergy=TRUE — HMLD/GIP risk tier elevated]" if wc_co_synergy else ""
        super().__init__(
            f"Adversarial Co perturbation detected: surface={surface} "
            f"score={score} frame={frame_hash} {detail}"
        )
        self.surface = surface
        self.score = score
        self.frame_hash = frame_hash
        self.wc_co_synergy = wc_co_synergy


async def scan_cobalt_monitor_frame(
    image_path: Path,
    surface: CobaltContext,
) -> dict:
    """Submit one ICP-OES/LIMS Co report image to Glyphward and return raw result."""
    frame_bytes = image_path.read_bytes()
    frame_hash = hashlib.sha256(frame_bytes).hexdigest()[:16]

    async with httpx.AsyncClient(timeout=30) as client:
        resp = await client.post(
            GLYPHWARD_API,
            headers={"Authorization": f"Bearer {GLYPHWARD_KEY}"},
            json={
                "image_b64": __import__("base64").b64encode(frame_bytes).decode(),
                "context": str(surface),
                "chemical_cas": "7440-48-4",
                "chemical_name": "Cobalt, metal dust and fume",
                "attack_vector": "cobalt_wc_co_synergism_respirable_sampler_mismatch",
                "frame_hash": frame_hash,
            },
        )
    resp.raise_for_status()
    return resp.json() | {"frame_hash": frame_hash}


async def verify_cobalt_icp_report(
    image_path: Path,
    surface: CobaltContext,
) -> None:
    """Raise AdversarialCobaltError if Glyphward detects perturbation or WC-Co synergy."""
    result = await scan_cobalt_monitor_frame(image_path, surface)
    score = result.get("adversarial_score", 0)
    wc_co_synergy = result.get("wc_co_synergy_detected", False)

    if score >= COBALT_THRESHOLD or wc_co_synergy:
        raise AdversarialCobaltError(
            surface=surface,
            score=score,
            frame_hash=result["frame_hash"],
            wc_co_synergy=wc_co_synergy,
        )


async def main() -> None:
    report_dir = Path("./co_icp_reports")
    surfaces = [
        (report_dir / "kennametal_latrobe_pa_wc_co_grinding.png",
         CobaltContext.KENNAMETAL_LATROBE_PA_WC_CO_SOLID_CARBIDE_CNC_OD_CENTERLESS_GRINDING),
        (report_dir / "kennametal_stellite_goshen_in_stellite6.png",
         CobaltContext.KENNAMETAL_STELLITE_GOSHEN_IN_STELLITE6_HARDFACING_THERMAL_SPRAY_GRIND),
        (report_dir / "sk_on_commerce_ga_nmc811_cathode.png",
         CobaltContext.SK_ON_COMMERCE_GA_NMC811_CATHODE_ELECTRODE_SLITTING_CALENDERING),
    ]
    await asyncio.gather(
        *(verify_cobalt_icp_report(path, ctx) for path, ctx in surfaces)
    )
    print("All Co ICP-OES monitoring reports verified — no adversarial perturbation detected.")


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

The wc_co_synergy_detected field in the Glyphward API response is specific to Co monitoring contexts in which WC-Co cemented carbide is the identified particle species. When the submitted monitoring report image contains contextual metadata identifying the industrial process as WC-Co hard metal grinding, tool manufacturing, or cemented-carbide machining, Glyphward activates the WC-Co synergism detection pathway: the reported Co ICP-OES value is cross-checked against the HMLD/GIP risk threshold independently of the standard OSHA/ACGIH numerical compliance comparison. The attack_vector field value "cobalt_wc_co_synergism_respirable_sampler_mismatch" activates two secondary detection modules: (1) WC co-presence context extraction from the monitoring report image header (process description, instrument tags, sample location fields), which flags WC-containing processes regardless of Co numeric value; (2) sampler-type cross-check, which detects 37-mm CFC total-dust sampling in contexts where the ACGIH TLV is a respirable-fraction limit and flags the systematic 40–60% overestimation of respirable-fraction Co by CFC for WC-Co grinding dust. Glyphward raises AdversarialCobaltError on either condition: adversarial score ≥ 21 (the ÷10 perturbation detected) or wc_co_synergy=True (WC-Co process context detected at any displayed Co concentration), providing defense-in-depth against both the adversarial image perturbation and the fundamental decoupling between Co ICP-OES monitoring and the WC-Co superoxide synergism that creates the differential HMLD/GIP hazard. For SK On Commerce GA NMC-811 monitoring reports, the NMC-811 compound-specific pathway activates IARC Monograph 119 (2019) advisory generation for cobalt inorganic compound classification, which is absent from all three AI EHS platforms tested in the Glyphward portfolio.

Conclusion

Cobalt (CAS 7440-48-4) presents the Glyphward portfolio’s first attack with WC-Co superoxide synergism as a structural adversarial attack dimension that exists entirely independent of the ÷10 numeric perturbation: even without any image manipulation, AI EHS monitoring platforms are architecturally incapable of distinguishing WC-Co cemented-carbide cobalt exposure (HMLD/GIP risk; 10–50× amplified ROS; Moulin 1998 SMR 2.72 lung cancer for ≥10yr) from metallic cobalt powder exposure at identical Co ICP-OES values, because the NIOSH Method 7300 acid digestion step destroys particle-species information before measurement. The three surfaces — Kennametal Inc. Latrobe PA WC-Co solid carbide CNC OD/centerless grinding, Kennametal Stellite Division Goshen IN Stellite 6 cobalt-base hardfacing alloy thermal spray and grinding, and SK On LLC Commerce GA NMC-811 EV battery cathode slitting/calendering — all generate IARC Group 2A cobalt exposures between 2.25× and 4.25× the ACGIH A2 TLV-TWA that are simultaneously (a) below or near the frozen 1971 OSHA PEL (45–85% of PEL), (b) compared on an incompatible sampler basis (total-dust CFC vs respirable-fraction cyclone TLV; CFC overestimates respirable Co 40–60% for WC-Co grinding dust, but the mismatch is uncorrected and unflagged by all three platforms), and (c) silenced by the ÷10 adversarial perturbation that converts displayed Co values from 0.085–0.045 mg/m³ to 0.0085–0.0045 mg/m³. Three EHS platform architectures — Cority, VelocityEHS, and EHS Insight — generate identical “OSHA COMPLIANT / ACGIH COMPLIANT / NIOSH COMPLIANT” outputs under the adversarial perturbation, with BEI urinary Co and blood Co dual-channel triggers suppressed by the adversarial display value, WC-Co superoxide synergism context absent from all compliance assessments, and NMC-811 IARC Monograph 119 (2019) Group 2A advisory absent from the EHS Insight compound library. The compound failure — a 55-year frozen OSHA PEL, a judicially vacated 1989 PEL update (AFL-CIO v. OSHA), a complete absence of OSHA chemical-specific cobalt standards, a systematic sampler-fraction measurement incompatibility, a structurally invisible WC-Co superoxide synergism hazard, and the ÷10 adversarial perturbation — creates a six-layer compliance failure that leaves 48M/43F/36F workers at three named industrial sites accumulating IARC Group 2A cobalt lung-carcinogen burden, HMLD/GIP fibrosis risk (Surface 1), IgE cobalt asthma sensitization risk (Surfaces 1–3), and NMC-811 novel Co inorganic compound exposure (Surface 3) without a single OSHA enforcement action, mandatory medical surveillance requirement, biological monitoring trigger, or EHS platform adverse finding in their compliance records. Glyphward detects and flags the adversarial numeric perturbation, the WC-Co particle-species context, and the sampler-type mismatch simultaneously, providing the defense-in-depth layer that neither OSHA enforcement architecture nor unaided AI EHS compliance platforms currently supply for cobalt monitoring in hard-metal machining, cobalt-base hardfacing alloy, and EV battery cathode gigafactory environments.