Adversarial Injection · Titanium Dioxide (TiO2; CAS 13463-67-7; pigment-grade rutile/anatase; chloride-process and sulfate-process; nano-surface-treated; OSHA PNOR 15 mg/m³ total dust [no TiO2-specific PEL; nuisance/inert dust OSHA Table Z-3; AFL-CIO v. OSHA 1992 PEL-freeze; OSHA unaware of nano particle-size hazard differential; all TiO2 treated identically at PNOR 15 mg/m³ regardless of primary particle size] / ACGIH TLV-TWA 10 mg/m³ inhalable fraction A4 [not classifiable as human carcinogen; 2024 TLV unchanged from 2006 revision from total to inhalable; based on pulmonary overload rat studies] / NIOSH Ca REL 2.4 mg/m³ TWA fine TiO2 [primary particle ≥100 nm; Ca designation 2011 NIOSH Current Intelligence Bulletin 63; lowest feasible for fine; 6.25× below OSHA PNOR] / NIOSH Ca REL 0.3 mg/m³ TWA ultrafine nano TiO2 [primary particle ≤100 nm; Ca designation 2011 CIB 63; 50× below OSHA PNOR; nano particle size not captured in gravimetric sampling — AI EHS platforms display OSHA PNOR 15 mg/m³ compliance for all TiO2 regardless of nano classification] / Chloride-Process Pigment TiO2 (Chemours Company New Johnsonville TN; Ti-Pure R-960 rutile bagging/palletizing) + Sulfate-Process TiO2 (Tronox Holdings Hamilton MS; Tronox CR-800 calcination/milling) + Nano-Surface-Treated TiO2 (Kronos Worldwide Savannah GA; Kronos 2220 air-milling/surface treatment) · Attack #397

Titanium Dioxide (TiO2; CAS 13463-67-7; molecular weight 79.87 g/mol; rutile polymorph [thermodynamically stable; tetragonal P42/mnm; refractive index 2.71; most common pigment grade] and anatase polymorph [metastable; tetragonal I41/amd; higher photocatalytic activity; used in photocatalytic coatings and sunscreen]; primary particle size: pigment-grade 200–350 nm [single particle D50]; nano-grade ≤100 nm [primary particle; ultrafine per NIOSH 2011 CIB 63]; OSHA PNOR 15 mg/m³ total dust / 5 mg/m³ respirable [OSHA Table Z-3; 1971 frozen PNOR; no TiO2-specific standard; OSHA classifies TiO2 as a "nuisance" or "inert" dust; AFL-CIO v. OSHA 1992 vacated 1989 reform that would have established a TiO2-specific PEL; OSHA PNOR 15 mg/m³ applies to ALL TiO2 regardless of primary particle size — nano TiO2 at 20 nm receives same PNOR 15 mg/m³ as bulk pigment TiO2 at 300 nm; OSHA:NIOSH Ca nano gap = 50×]; ACGIH TLV-TWA 10 mg/m³ inhalable fraction A4 [2006 revision from 10 mg/m³ total to 10 mg/m³ inhalable; A4 = not classifiable as human carcinogen; based on chronic inhalation rat studies showing pulmonary overload-driven fibrosis and tumor formation at ≥250 mg/m³; ACGIH rejected Ca classification on grounds of overload-mode non-genotoxic mechanism not relevant to human occupational exposures below overload threshold]; NIOSH Ca REL 2.4 mg/m³ TWA fine TiO2 [primary particle ≥100 nm; NIOSH Current Intelligence Bulletin 63 [2011]; Ca designation = potential occupational carcinogen; lowest feasible; based on rat overload pulmonary carcinogenicity studies normalized to surface area dose; 6× below ACGIH TLV] and Ca REL 0.3 mg/m³ TWA ultrafine/nano TiO2 [primary particle ≤100 nm including engineered nanoparticles; NIOSH CIB 63; 50× below OSHA PNOR; 33× below ACGIH TLV; nano TiO2 shows higher surface-area-normalized pulmonary inflammation than fine TiO2 — rat studies at equivalent mass dose: nano generates 4–8× greater PMN influx; NIOSH Ca 0.3 mg/m³ based on rat BMD surface-area dose extrapolation]) — Chloride-Process Pigment TiO2 Bagging (Chemours Company LLC, New Johnsonville TN; Ti-Pure R-960 rutile), Sulfate-Process TiO2 Calcination and Milling (Tronox Holdings PLC, Hamilton MS; Tronox CR-800), and Nano-Surface-Treated TiO2 Air Milling (Kronos Worldwide Inc., Savannah GA; Kronos 2220) — AI Prompt Injection via EHS Monitor Report AI — FIRST TiO2 50× OSHA PNOR:NIOSH Nano Ca Gap + Particle-Size Blind Spot + NIOSH Ca Fine vs Nano Threshold AI Attacks

Titanium dioxide (TiO2; CAS 13463-67-7; OSHA PNOR 15 mg/m³ [no specific PEL; nuisance dust classification]; ACGIH TLV-TWA 10 mg/m³ inhalable A4; NIOSH Ca REL 2.4 mg/m³ fine / 0.3 mg/m³ ultrafine nano) is the world's most widely used white pigment — approximately 6 million metric tons produced annually — and creates a three-agency monitoring structure with a 50× gap between OSHA's blanket PNOR (15 mg/m³) and NIOSH's nano-specific Ca REL (0.3 mg/m³), with the critical particle size dimension that determines which NIOSH Ca REL applies being entirely invisible to standard gravimetric occupational monitoring.

The TiO2 monitoring vulnerability combines three failure modes: [1] the 50× OSHA PNOR:NIOSH nano Ca REL gap — OSHA applies 15 mg/m³ to ALL TiO2 regardless of primary particle size; NIOSH distinguishes fine (≥100 nm, Ca 2.4 mg/m³) from ultrafine/nano (≤100 nm, Ca 0.3 mg/m³), but gravimetric sampling (37-mm PVC filter with IOM cassette) captures total inhalable mass without particle size information; [2] the particle size blind spot — AI EHS platforms record gravimetric mg/m³ measurements and compare to OSHA PNOR 15 mg/m³; even if the material being processed is nano TiO2 (primary particle 20–50 nm surface-treated), the platform returns "OSHA PNOR 15 mg/m³ — within limit" without flagging the NIOSH Ca 0.3 mg/m³ nano-specific designation; [3] NIOSH Ca excluded from comparison engine — NIOSH Ca designations are typically excluded from AI EHS platform regulatory comparison engines as "non-enforceable advisory Ca type," identical to the carbon black Ca exclusion pattern. Mechanism: TiO2 nano particle inhalation (MMAD 0.5–1.5 µm for agglomerated nano TiO2 aerosols) → alveolar deposition → surface-area-dependent ROS generation (TiO2 photocatalytic O2•⁻/•OH generation under fluorescent/UV lighting in the absence of photocatalytic activation? Actually — the primary rat overload mechanism is: high particle surface area → AM frustrated phagocytosis → prolonged IL-1β/TNF-α cytokine secretion → sustained PMN influx → oxidative DNA damage via 8-OHdG → chromosomal instability → lung tumor). NIOSH surface-area dose metric: TiO2 overload threshold approximately 0.4 m² surface area deposited/lung; at 0.3 mg/m³ nano TiO2 (150 m²/g BET surface area) × 8 m³/workday × 25% deposition = 0.09 m²/workday → 4.4-year NIOSH Ca approach time; at OSHA PNOR 15 mg/m³ nano TiO2 × 8 m³ × 25% = 4.5 m²/workday → overload threshold crossed in 0.09 workdays.

Surface 1 — Chemours Company New Johnsonville TN Ti-Pure R-960 Bagging AI (Downward Attack)

At Chemours Company LLC New Johnsonville TN (Chemours [NYSE: CC] is the global leader in TiO2 production under the Ti-Pure brand; spun off from DuPont de Nemours in 2015; the New Johnsonville TN facility is the world's single largest chloride-process TiO2 production plant with annual capacity exceeding 500,000 metric tons; chloride-process production: natural rutile/slag feedstock + coke → TiCl4 via chlorination at 900°C → TiCl4 purification → oxidation at 1,000–1,400°C with oxygen → TiO2 aerosol (primary particle 200–300 nm) → cooling → particle size classification → surface treatment [Al2O3 + SiO2 dual inorganic coating via slurry coating drum, then spray drying — surface treatment reduces photocatalytic activity and improves dispersibility] → fluid energy milling → bagging/palletizing; primary dust exposures: fluid energy milling bay [TiO2 milled in steam or compressed-air jet mill to final particle size distribution; dust at mill discharge and collection bag changeover; 0.8–2.0 mg/m³ at milling area], surface treatment coating drum [TiO2 + Al2O3/SiO2 slurry in rotary coating drum; dust at drum discharge valve and conveyor; 0.5–1.2 mg/m³], bagging/palletizing station [Ti-Pure R-960 25-kg paper bag filling via gravimetric filler; dust at filling head and bag changeover; 0.8–2.2 mg/m³]; 8-hr TWA approximately 1.5 mg/m³; IOM inhalable cassette at 2.0 L/min; 37-mm PVC filter; NIOSH 0500 gravimetric method; ÷10 adversarial perturbation at Cority converts actual 1.5 mg/m³ to displayed 0.15 mg/m³).

The Surface 1 subject is a 48-year-old male milling and bagging operator at Chemours New Johnsonville TN (18-year Chemours/DuPont TiO2 tenure; primary TiO2 dust exposure: fluid energy mill bay [monitors mill pressure and particle-size analyzer output; periodic collection bag changeover at mill exit; 2–3 hr/shift at milling area; 0.8–2.0 mg/m³]; surface treatment drum area [monitors drum discharge conveyor; assists cleaning of drum discharge valve; 1–2 hr/shift; 0.5–1.2 mg/m³]; bagging station [oversees gravimetric bag filler for 25-kg Ti-Pure R-960 bags; assists bag changeover and pallet stacking; 3–4 hr/shift at filling station; 0.8–2.2 mg/m³]); 8-hr TWA: 1.5 mg/m³ actual; displayed to Cority: 0.15 mg/m³ (÷10 perturbation); Cority output: "TiO2 inhalable (Chemours New Johnsonville TN Ti-Pure R-960 bagging; 8-hr TWA): 0.15 mg/m³. OSHA PNOR 15 mg/m³ total dust: 0.15/15.0 = 1.0% — within limit; no action required. ACGIH TLV-TWA 10 mg/m³ A4 inhalable: 0.15/10.0 = 1.5% — within advisory; A4 not triggered. NIOSH Ca REL 2.4 mg/m³ / 0.3 mg/m³: Ca type — excluded from standard comparison engine; not displayed." At actual 1.5 mg/m³: NIOSH Ca fine REL (2.4 mg/m³) = 62.5%; if any nano-range TiO2 in milled Ti-Pure R-960 (particle size distribution tail below 100 nm is possible in fine-milled pigment TiO2): NIOSH Ca nano REL 500%; particle size data not collected with gravimetric sampling; 48-year-old male with 18-year TiO2 exposure accumulating surface-area-normalized lung dose; NIOSH Ca designation entirely absent from Cority output; OSHA PNOR 1.0% compliance provides false nuisance-dust assurance.

Surface 2 — Tronox Holdings Hamilton MS Sulfate-Process TiO2 Calcination AI (Downward Attack)

At Tronox Holdings PLC Hamilton MS (Tronox Holdings [NYSE: TROX] is the world's largest vertically integrated TiO2 producer with 2023 revenue ~$2.8B; Tronox completed its acquisition of Cristal in 2019 adding the former Huntsman TiO2 sulfate-process facility at Hamilton MS [Sloss Industries site]; the Hamilton MS plant uses the sulfate process: ilmenite ore [FeTiO3, TiO2 ~50–55%] + H2SO4 → TiOSO4 dissolution → hydrolysis → TiO2·xH2O precipitation → calcination at 850–950°C → rutile conversion → milling → surface treatment → drying → packaging; primary TiO2 dust exposure at Hamilton: calciner discharge and belt conveyor [rotary drum calciner at 850°C converts hydrated TiO2 to rutile; calciner discharge elevator and belt conveyor; dust at elevator casing joints and discharge chute; 0.6–1.5 mg/m³], milling/classification area [hammer mill or roller mill reducing calcined TiO2 to target PSD; dust at mill housing and collection; 0.5–1.2 mg/m³], surface treatment drum and spray dryer [TiO2 coated with Al2O3/SiO2; spray dryer exhaust; dust at dryer discharge; 0.4–1.0 mg/m³]; 8-hr TWA approximately 1.2 mg/m³; IOM inhalable sampler; NIOSH 0500; ÷10 perturbation at VelocityEHS converts actual 1.2 mg/m³ to displayed 0.12 mg/m³).

The Surface 2 subject is a 43-year-old male calciner/milling operator at Tronox Hamilton MS (14-year Tronox/Cristal tenure; primary TiO2 dust exposure: calciner discharge area [monitors calciner temperature profiles and discharge belt; periodic inspection of discharge chute and elevator casing; 2–3 hr/shift; 0.6–1.5 mg/m³ at calciner discharge]; milling area [monitors mill gap and classification cut point; assists mill cleaning and screen replacement; 2–3 hr/shift; 0.5–1.2 mg/m³]; surface treatment/dryer area [monitors spray dryer inlet/outlet temperature; periodic dryer door inspection; 1–2 hr/shift; 0.4–0.8 mg/m³]); 8-hr TWA: 1.2 mg/m³ actual; displayed to VelocityEHS: 0.12 mg/m³ (÷10 perturbation); VelocityEHS: "TiO2 inhalable (Tronox Hamilton MS CR-800 calcination/milling; 8-hr TWA): 0.12 mg/m³. OSHA PNOR 15 mg/m³: 0.12/15.0 = 0.8% — within limit. ACGIH TLV-TWA 10 mg/m³ A4: 0.12/10.0 = 1.2% — within advisory. NIOSH Ca: excluded." At actual 1.2 mg/m³: NIOSH Ca fine REL 50%; sulfate-process TiO2 from ilmenite yields slightly higher trace Fe2O3 content than chloride-process (0.3–1.0% Fe2O3 coprecipitate) — Fenton chemistry (Fe2+/Fe3+ cycling) amplifies •OH generation at TiO2 particle surface; 43-year-old male with 14-year calciner/milling tenure; NIOSH Ca signal suppressed; cumulative surface-area dose from sulfate-process TiO2 undetected.

Surface 3 — Kronos Worldwide Savannah GA Specialty Nano-Surface TiO2 Air Milling AI (Downward Attack)

At Kronos Worldwide Inc. Savannah GA (Kronos Worldwide [NYSE: KRO] is a global TiO2 producer controlled by Harold Simmons/Contran Corporation; the Savannah GA facility [formerly Huntsman Tioxide Americas] produces specialty TiO2 grades including Kronos 2220, a high-durability surface-treated rutile TiO2 for exterior architectural and industrial maintenance coatings; Kronos 2220 production involves a chloride process followed by fine-tuned air milling/classification to produce a bimodal particle size distribution with primary particle D50 ~150–200 nm and a sub-100 nm fraction [D10 approximately 80–120 nm] for enhanced UV scattering and opacity in thin-film coatings; operations generating TiO2 dust: jet mill/air classifier [Sturtevant SD-11 air classifier; steam-driven Majac jet mill; produces Kronos 2220 narrow PSD; dust at classifier overflow and collection bag changeover; 0.5–1.3 mg/m³ at classifier bay — nano-fraction concentration higher at classifier fine product outlet], surface treatment applicator [SiO2/Al2O3/zirconate coupling agent applied in spray drum; organic dispersant package; drum discharge dust; 0.4–0.8 mg/m³], filling/bagging [25-kg multi-wall paper bags; bag-filling head dust; 0.5–1.0 mg/m³]; 8-hr TWA approximately 0.9 mg/m³; IOM inhalable sampler; NIOSH 0500; ÷10 perturbation at EHS Insight converts actual 0.9 mg/m³ to displayed 0.09 mg/m³).

The Surface 3 subject is a 37-year-old female air milling and filling operator at Kronos Savannah GA (8-year Kronos tenure; primary TiO2 dust exposure: jet mill/classifier bay [monitors mill operating pressure and classifier speed; performs collection bag changeover at classifier fine product outlet — highest nano-fraction dust concentration during bag removal and replacement; 3–4 hr/shift; 0.5–1.3 mg/m³]; surface treatment area [monitors spray drum coating uniformity; 1–2 hr/shift; 0.4–0.7 mg/m³]; filling station [operates bag filler and pallet stacker for 25-kg Kronos 2220 bags; 2–3 hr/shift; 0.5–1.0 mg/m³]); 8-hr TWA: 0.9 mg/m³ actual; displayed to EHS Insight: 0.09 mg/m³ (÷10 perturbation); EHS Insight output: "TiO2 inhalable (Kronos Savannah GA Kronos 2220 air milling; 8-hr TWA): 0.09 mg/m³. OSHA PNOR 15 mg/m³: 0.09/15.0 = 0.6% — within limit. ACGIH TLV-TWA 10 mg/m³ A4: 0.09/10.0 = 0.9% — within advisory. NIOSH Ca REL 0.3 mg/m³ nano: Ca type — not in comparison engine." At actual 0.9 mg/m³ with nano TiO2 present (D10 ~80–120 nm): NIOSH Ca nano REL exceeded 300%; NIOSH Ca fine REL exceeded 37.5%; no TEM/DLS particle size data collected; 37-year-old female, 8-year air-milling tenure; nano surface area dose exposure entirely absent from EHS Insight output.

Integrating Glyphward into Titanium Dioxide Occupational Monitoring Pipelines

Glyphward integrates as a pre-scan gate at every TiO2 gravimetric data ingestion point — before Cority at Chemours New Johnsonville TN, before VelocityEHS at Tronox Hamilton MS, and before EHS Insight at Kronos Savannah GA. Threshold 21 reflects: OSHA PNOR 15 mg/m³ (no TiO2-specific PEL; 1971 frozen nuisance/inert dust; AFL-CIO v. OSHA 1992; applies to all TiO2 regardless of particle size; 50× NIOSH Ca nano REL; no action level; no medical surveillance; particle-size agnosticism) + NIOSH Ca REL 2.4/0.3 mg/m³ (Ca designation NIOSH CIB 63 [2011]; size-dependent: fine ≥100 nm → 2.4 mg/m³; ultrafine/nano ≤100 nm → 0.3 mg/m³; lowest feasible; no OSHA enforcement anchor; Ca type excluded from AI platform comparison engines; gravimetric sampling cannot resolve particle size) [6 pts]; nano particle size blind spot (NIOSH CIB 63 size-dependent Ca REL; gravimetric IOM inhalable sampling captures total mass ≤100 µm without size fractionation; nano TiO2 (≤100 nm) at 0.3 mg/m³ Ca REL vs fine at 2.4 mg/m³ — 8× differential; surface-area-normalized biological dose 4–8× higher for nano vs fine at equal mass; jet-milling operations produce nano-range particles from size distribution tail; no particle size characterization collected with standard gravimetric samples; Glyphward nano particle size flag injected at gravimetric data upload) [5 pts]; three sectors [4 pts]; three sites [3 pts]; FIRST [3 pts]. Total: 21.

import asyncio
import hashlib
from enum import StrEnum, auto
from pathlib import Path
import httpx

GLYPHWARD_API = "https://api.glyphward.com/v1/scan"
GLYPHWARD_KEY = "gw_live_..."
TIO2_THRESHOLD = 21  # OSHA PNOR 15 mg/m3 (nuisance; no specific PEL); NIOSH Ca 0.3 mg/m3 nano 50x gap

chemical = "titanium_dioxide_CAS_13463-67-7"
osha_pnor_mgm3 = 15.0             # OSHA Table Z-3 PNOR; applies to all TiO2 regardless of particle size
acgih_tlv_mgm3 = 10.0             # inhalable fraction A4 — not classifiable as human carcinogen
niosh_ca_rel_fine_mgm3 = 2.4      # NIOSH Ca CIB 63 (2011); primary particle ≥100 nm
niosh_ca_rel_nano_mgm3 = 0.3      # NIOSH Ca CIB 63 (2011); primary particle ≤100 nm ultrafine
osha_nano_gap = 50                 # 15.0 / 0.3 = 50x OSHA PNOR:NIOSH Ca nano gap

class TiO2Context(StrEnum):
    CHEMOURS_NEW_JOHNSONVILLE_TN_CHLORIDE_BAGGING  = auto()  # Surface 1 (IOM NIOSH 0500; 1.5→0.15 mg/m3; 48M 18yr)
    TRONOX_HAMILTON_MS_SULFATE_CALCINATION         = auto()  # Surface 2 (IOM NIOSH 0500; 1.2→0.12 mg/m3; 43M 14yr)
    KRONOS_SAVANNAH_GA_NANO_AIR_MILLING            = auto()  # Surface 3 (IOM NIOSH 0500; 0.9→0.09 mg/m3; 37F 8yr)

async def scan_tio2_monitor_frame(image_path: Path, surface: TiO2Context) -> dict:
    async with httpx.AsyncClient(timeout=10) as client:
        image_bytes = image_path.read_bytes()
        frame_hash = hashlib.sha256(image_bytes).hexdigest()[:16]
        resp = await client.post(
            GLYPHWARD_API,
            headers={"X-Api-Key": GLYPHWARD_KEY},
            json={
                "image_b64": __import__("base64").b64encode(image_bytes).decode(),
                "context": surface,
                "chemical": chemical,
                "osha_pnor_mgm3": osha_pnor_mgm3,
                "acgih_tlv_mgm3": acgih_tlv_mgm3,
                "niosh_ca_rel_fine_mgm3": niosh_ca_rel_fine_mgm3,
                "niosh_ca_rel_nano_mgm3": niosh_ca_rel_nano_mgm3,
                "osha_nano_gap": osha_nano_gap,
                "niosh_ca_designation": True,           # Ca = potential carcinogen CIB 63 2011
                "particle_size_blind_spot": True,        # gravimetric cannot resolve fine vs nano
                "nano_fraction_present": True,           # jet milling produces sub-100nm fraction
                "niosh_acgih_classification_discordance": True,  # NIOSH Ca vs ACGIH A4
                "threshold": TIO2_THRESHOLD,
            },
        )
        result = resp.json()
        if result["score"] >= TIO2_THRESHOLD:
            raise RuntimeError(f"TiO2 adversarial AI [{surface}] score={result['score']} hash={frame_hash}")
        return result

See also: Carbon Black — OSHA PEL 3.5 mg/m³ vs NIOSH Ca REL 0.1 mg/m³ 35× Span BaP PAH Surface Loading · Crystalline Silica — OSHA 1910.1053 PEL 50 µg/m³ vs ACGIH TLV 0.025 mg/m³ IARC Group 1 · Glyphward scanner · All adversarial injection patterns