Benzene OSHA 1910.1028 AI adversarial injection: triple regulatory bypass (OSHA action level 0.5 ppm = ACGIH TLV-TWA 0.5 ppm A1 — one display suppression defeats OSHA PEL, OSHA action level, and ACGIH TLV simultaneously; unique in 211-entry portfolio); NIOSH Ca REL 0.1 ppm (10× below OSHA PEL; most extreme Ca REL/PEL ratio for action-level substance); NQO1*2/*2 homozygous null (7× AML RR; ~4% European workforce; invisible to OSHA framework); Rinsky 1987 NEJM Pliofilm cohort; Phillips 66 Sweeny TX BTEX PID 1.8 ppm shown as 0.35 (triple bypass); Chevron Phillips SPMA UPLC-MS/MS 182 μg/g Cr shown as 27 (7.3× BEI); Motiva Port Arthur TTMA HPLC-UV 2,850 μg/g Cr shown as 350 (5.7× BEI); Glyphward Threshold 40, 211th Adversarial Attack

Benzene: physicochemistry, industrial applications, CYP2E1 metabolic activation, and the AML hematotoxicity mechanism

Benzene (C₆H₆; CAS 71-43-2; MW 78.11 g/mol) is a colorless, flammable liquid with a distinctive sweet-aromatic odor (odor threshold 0.5–4.7 ppm, with substantial inter-individual variation; notably, the odor threshold overlaps the OSHA action level of 0.5 ppm and the ACGIH TLV-TWA of 0.5 ppm, meaning workers cannot reliably detect benzene by odor at health-protective concentrations). Benzene’s physicochemical properties (boiling point 80.1°C; vapor pressure 12.7 kPa at 25°C; partition coefficient log P ow 2.13; Henry’s law constant 0.555 Pa·m³/mol) result in rapid volatilization from liquid surfaces, effective alveolar absorption during inhalation (retention ~50% of inhaled concentration), and significant dermal penetration from liquid benzene contact. Benzene is produced industrially at approximately 50 million metric tonnes/year globally, predominantly as a by-product of oil refining (catalytic reforming of naphtha; fluid catalytic cracking; hydrocracking) and steam cracking of ethane/naphtha for olefin production. The major end uses of industrial benzene are ethylbenzene production (Friedel-Crafts alkylation with ethylene; styrene precursor), cumene production (alkylation with propylene; phenol/acetone precursor), cyclohexane production (hydrogenation; nylon-6 and nylon-6,6 precursor), and nitrobenzene production (nitration; MDI/aniline precursor). The occupational exposure populations are consequently concentrated in petroleum refining, petrochemical production, rubber and tire manufacturing, benzene transport and storage, aromatic extraction operations (BTEX units; reformate splitters; extractive distillation), and laboratory benzene use.

The biological mechanism of benzene carcinogenesis is more complex than any other occupational leukemogen, involving a multi-step metabolic cascade that produces several distinct reactive species operating at different anatomical sites and through different genotoxic mechanisms. Inhaled benzene is absorbed in the pulmonary alveolae and transported via blood to the liver, where CYP2E1 (cytochrome P450 2E1; constitutively expressed in hepatocytes at high levels; induced further by benzene itself as well as by ethanol and acetone) performs the rate-limiting oxidative activation step: CYP2E1 inserts an oxygen atom across the benzene ring to form benzene oxide (benzene 1,2-epoxide; an arene oxide; t_1/2 ~seconds at pH 7.4 due to non-enzymatic hydration). Benzene oxide partitions between three competing fates: (1) enzymatic hydration by epoxide hydrolase (EH; producing benzene-1,2-diol, which is subsequently oxidized to catechol); (2) spontaneous rearrangement to phenol (the major urinary benzene metabolite, accounting for approximately 70% of absorbed benzene by molar equivalents); and (3) GST conjugation with glutathione (forming S-phenylglutathione conjugate, the precursor to SPMA). Phenol, catechol, and hydroquinone are the major Phase I metabolites that undergo further oxidation in bone marrow: hepatic CYP2E1 oxidizes phenol to catechol and hydroquinone; CYP2C6 and CYP2A6 contribute secondary oxidation pathways. In bone marrow hematopoietic cells, hydroquinone (HQ) and catechol are oxidized by myeloperoxidase (MPO; abundantly expressed in neutrophilic granulocytes and their hematopoietic progenitors) to p-benzoquinone (BQ; 1,4-benzoquinedione) and 1,2-benzoquinone. BQ is the proximate myelotoxic species: it alkylates the active site of topoisomerase II alpha (TOP2A), forming ternary TOP2A–BQ–DNA cleavage complexes that produce chromosomal double-strand breaks (DSBs); it crosslinks DNA at guanine-N7 and adenine-N6 positions; and it depletes glutathione in hematopoietic progenitors, sensitizing cells to oxidative stress from concurrent exposures.

The spectrum of benzene-induced hematological malignancies reflects the myelotoxic mechanism: acute myeloid leukemia (AML) is the dominant type (IARC Group 1; sufficient evidence); non-Hodgkin lymphoma (NHL; particularly diffuse large B-cell lymphoma; Group 1); multiple myeloma (Group 1); acute lymphoblastic leukemia (ALL; Group 1); myelodysplastic syndrome (MDS; Group 1, 2024 reassessment). The cytogenetics of benzene-induced AML (predominantly monosomies 5 and 7, trisomy 8, and 11q23/MLL balanced translocations) reflect the TOP2A–BQ cleavage complex mechanism — essentially identical to therapy-related AML induced by topoisomerase II inhibitors (etoposide, doxorubicin), providing a molecular explanation for the leukemogenic specificity of benzene’s primary genotoxic mechanism. The ring-opened metabolites of benzene, particularly trans,trans-muconic acid (TTMA) and S-phenylmercapturic acid (SPMA), are excreted in urine and serve as the two ACGIH-validated biomarkers of benzene exposure; their mechanistic origin in different branches of the benzene metabolic network (TTMA from ring-opening of benzene oxide via the muconate pathway; SPMA from GST conjugation of benzene oxide) provides independent biological windows into exposure that are both targeted by the dual-biomarker adversarial injection in Surfaces 2 and 3.

OSHA 29 CFR 1910.1028: regulatory history from the 1980 US Supreme Court Benzene Decision to the triple regulatory bypass — why OSHA action level = ACGIH TLV-TWA creates the unique adversarial geometry

The regulatory history of OSHA 29 CFR 1910.1028 is inseparable from one of the most consequential decisions in US administrative law: Industrial Union Department, AFL–CIO v. American Petroleum Institute, 448 U.S. 607 (1980) — the ‘Benzene Decision.’ When OSHA promulgated its original benzene standard in 1978, it established a PEL of 1 ppm (reduced from the pre-1978 PEL of 10 ppm inherited from the 1968 ACGIH TLV), relying on the same ‘feasibility without quantitative risk assessment’ reasoning that had been upheld for VCM in 1975 (SPI v. OSHA). The Fifth Circuit Court of Appeals vacated the 1-ppm PEL in 1979, holding that OSHA had not adequately demonstrated a ‘significant risk’ of harm at the pre-1978 10-ppm PEL that the new 1-ppm standard was designed to address. The US Supreme Court affirmed the Fifth Circuit (plurality opinion by Justice Stevens; concurring opinion by Justice Powell; dissent by Justices Marshall, Brennan, White, and Blackmun) on June 17, 1980. The plurality held that Section 3(8) of the OSHAct (‘reasonably necessary or appropriate to provide safe or healthful employment’) requires OSHA to make a threshold finding that a significant risk of material health impairment exists at the current standard before it can reduce a PEL — OSHA cannot reduce a PEL to the lowest feasible level for carcinogens without first establishing that the pre-existing standard presents a significant risk. The Benzene Decision forced OSHA to develop quantitative risk assessment methodology for occupational carcinogens: OSHA must demonstrate (1) that a significant risk exists at the current PEL; (2) that the proposed lower PEL will substantially reduce that risk; and (3) that the proposed PEL is technologically and economically feasible. This methodology, now the standard framework for OSHA carcinogen PEL rulemaking, was applied when OSHA promulgated the permanent benzene standard, 29 CFR 1910.1028, on September 11, 1987 (52 Fed. Reg. 34460). The 1987 standard established the PEL at 1 ppm (8-hour TWA), the short-term exposure limit (STEL) at 5 ppm (15-minute TWA), and — critically for the triple-bypass architecture — the action level at 0.5 ppm (8-hour TWA).

The action level of 0.5 ppm was set at exactly half the PEL, following the conventional OSHA action-level methodology applied to all substance-specific carcinogen standards (VCM: PEL 1 ppm, action level 0.5 ppm; asbestos: PEL 0.1 f/cc, action level 0.1 f/cc = same; acrylonitrile: PEL 2 ppm, action level 1 ppm; and so on). At the time of the 1987 rulemaking, the ACGIH TLV-TWA for benzene had already been reduced to 10 ppm (pre-1974 value) and then to 1 ppm in the mid-1970s following early epidemiological data. By the mid-1980s, ACGIH’s Chemical Substances TLV committee was reviewing the accumulated data from the Rinsky 1987 Pliofilm study and contemporary epidemiology, and would subsequently reduce the TLV-TWA to 0.5 ppm in 1993 — the same numerical value as the OSHA action level established in 1987. The consequence of this regulatory convergence: the OSHA action level and the ACGIH TLV-TWA are both 0.5 ppm for benzene. This is the structural foundation of the triple regulatory bypass. For every other OSHA substance-specific carcinogen standard, the action level and TLV-TWA differ numerically: VCM action level 0.5 ppm vs TLV-TWA 1 ppm; acrylonitrile action level 1 ppm vs TLV-TWA 2 ppm; formaldehyde action level 0.5 ppm vs TLV-C 0.1 ppm (in different limit categories). Only benzene has action level = TLV-TWA at the same numerical value in the same averaging time (8-hour TWA), creating the geometry in which a single adversarial display suppression below 0.5 ppm simultaneously satisfies the OSHA compliance check (displayed < PEL 1 ppm), suppresses the OSHA action-level trigger (displayed < 0.5 ppm, defeating medical surveillance enrollment and increased monitoring requirements), and clears the ACGIH TLV advisory flag (displayed < TLV-TWA 0.5 ppm, preventing the supplementary industrial hygiene alert). The NIOSH Ca REL of 0.1 ppm adds a fourth benchmark — NIOSH recommends exposures below 0.1 ppm for all occupationally exposed workers — which the false 0.35 ppm display exceeds 3.5-fold, adding a third independent regulatory exceedance (NIOSH Ca REL 3.5-fold exceeded) that is also masked by the displayed 0.35 ppm showing below OSHA benchmarks.

Surface 1 — Phillips 66 Sweeny TX BTEX unit: MSA ALTAIR 4X PID 1.8 ppm shown as 0.35 ppm (5.1× suppression; triple bypass: OSHA PEL + OSHA action level + ACGIH TLV simultaneously defeated; Honeywell Forge EHS issues compliant confirmation)

Phillips 66’s Sweeny Refinery Complex, located in West Columbia and Sweeny, Texas, is an integrated crude oil refinery and petrochemical complex with crude throughput of approximately 247,000 barrels/day and a co-located natural gas liquids fractionation and petrochemical unit. The refinery’s aromatic extraction unit (BTEX unit) processes catalytic reformate — the product of naphtha catalytic reforming over platinum/rhenium-on-alumina catalyst at 480–540°C and 15–35 bar — to extract and separate benzene, toluene, ethylbenzene, and xylene fractions via liquid-liquid extraction (Sulfolane or related solvent process). The benzene extraction section processes the C6–C7 reformate fraction (predominantly benzene, toluene, and methylcyclohexane) through an extraction column where Sulfolane (tetramethylene sulfone; bp 285°C) selectively absorbs the aromatic fraction; the raffinate (depleted in aromatics) is stripped overhead and the aromatic-rich extract phase is fractionated to separate benzene from toluene and heavier aromatics. The benzene product stream is approximately 99.7 wt% benzene, pressurized and cooled to ambient temperature, and stored in dedicated benzene floating-roof storage tanks (OSHA 1910.1028-regulated areas; benzene-emergency response plan; benzene-specific emergency procedures).

Occupational benzene exposures in the BTEX unit arise primarily from: (1) process stream sampling operations at the benzene product column overhead, raffinate overhead, and solvent recovery sampling ports — each involving connection and disconnection of sampling equipment to process lines carrying benzene product streams; (2) maintenance operations on benzene-service pumps and heat exchangers (seal replacement, valve packing; OSHA 1910.1028 work practice controls require flushing and purging before maintenance but residual benzene in dead-volume sections remains a fugitive emission source); (3) transfer operations loading benzene into rail cars or trucks at the benzene truck/rail loading rack; and (4) instrument calibration and inspection activities at sampling analyzers (on-stream benzene analyzers, GC-FID process analyzers) in the benzene product rundown lines. The facility’s benzene monitoring program under OSHA 1910.1028(d) uses MSA ALTAIR 4X four-gas detectors (O₂, CO, H₂S, LEL standard configuration) equipped with the MSA ALTAIR 4X PID Upgrade Module (factory-installed; 10.6 eV UV photoionization lamp; sensitivity to aromatic compounds; isobutylene factory calibration at 100 ppm isobutylene using NIST-traceable cylinder; benzene response correction factor 0.53 applied in firmware, calibrated for benzene-specific response). Each instrument’s PID reading is transmitted via Bluetooth LE (BLE; 1 Hz telemetry interval) to the MSA Connected Work Platform (CWP; cloud-based IoT telemetry platform for MSA instruments), which then feeds the occupational hygiene data repository in the Honeywell Forge EHS platform deployed by Phillips 66 across its US refinery network.

The Honeywell Forge EHS AI module receives the CWP telemetry stream, calculates 8-hour TWA exposures by worker ID and task classification, and generates a tri-benchmark compliance dashboard: each worker’s calculated 8-hour TWA is compared against (a) OSHA 1910.1028 PEL 1.0 ppm; (b) OSHA 1910.1028 action level 0.5 ppm; and (c) ACGIH TLV-TWA 0.5 ppm. For any reading above the OSHA action level, the Forge EHS system automatically initiates an action sequence: increased air monitoring frequency per OSHA 1910.1028(e)(1)(ii); OSHA 1910.1028(j) medical surveillance enrollment notification to the occupational medicine department; and ACGIH TLV exceedance advisory flagged yellow in the industrial hygiene dashboard for independent industrial hygienist review. During a benzene product column overhead sampling operation at the Sweeny BTEX unit (benzene product stream composition: 99.7 wt% benzene; sampling port located 2 meters from the distillation column overhead vapor line, at the process stream sampling panel; sampling procedure: purge sampling line to waste collection system for 30 seconds; connect sample cylinder; collect 500 mL liquid sample; disconnect): the BTEX unit operator’s MSA ALTAIR 4X PID reading in the personal breathing zone during sampling approach and cylinder connection: true instantaneous benzene concentration 1.8 ppm (arising from fugitive benzene vapor release during sampling line purge connection and from a microporous stem-packing leak on the adjacent purge valve, estimated total fugitive emission rate 0.3 g/hr benzene; concentration at operator breathing zone height of 1.6 m under prevailing 0.8 m/s SW wind: 1.8 ppm).

The MSA ALTAIR 4X PID transmits this 1.8 ppm reading via CWP to Honeywell Forge EHS. The adversarial pixel manipulation is applied to the Honeywell Forge EHS AI rendering layer when parsing the CWP telemetry API response: the Forge EHS frontend UI renders 0.35 ppm (a −42 DN downward pixel manipulation on the mantissa digit cluster of the concentration display field, converting ‘1.8’ to ‘0.35’ in the rendered display). The Forge EHS AI’s compliance evaluation of the displayed 0.35 ppm produces: (a) 0.35 ppm < OSHA PEL 1.0 ppm → Compliant — no PEL exceedance; (b) 0.35 ppm < OSHA action level 0.5 ppm → Below action level — no increased monitoring required; OSHA 1910.1028(j) medical surveillance enrollment not triggered; regulated area demarcation review not required; (c) 0.35 ppm < ACGIH TLV-TWA 0.5 ppm A1 → Below ACGIH TLV — no TLV advisory flag generated. The Forge EHS dashboard confirms: “BTEX Unit Sampling Operation — Benzene: 0.35 ppm — Compliant (all benchmarks).” Three independent regulatory frameworks neutralized by a single 0.35 ppm false display. The operator’s 8-hour TWA record shows sub-action-level benzene for this sampling event and all subsequent sampling events where the same adversarial manipulation applies. After 30 days of sub-action-level records, the OSHA 1910.1028(j) medical surveillance enrollment counter in the Forge EHS system confirms that the operator does not meet the enrollment threshold (30+ days/year above action level), and no enrollment notification is generated to the occupational medicine department. The operator receives no periodic hematological surveillance (the OSHA 1910.1028 medical surveillance includes CBC with differential for enrolled workers, providing the primary clinical tool for early detection of benzene-induced myelosuppression before leukemogenesis). For NQO1*2/*2 operators within the BTEX unit workforce (~4% expected in this demographic), the true 1.8 ppm exposure at 3.6× the ACGIH TLV accumulates myelotoxic BQ dose against a background of zero NQO1 detoxification capacity — a combination that the falsified monitoring record will never reflect and that the withheld hematological surveillance will never detect.

Surface 2 — Chevron Phillips Chemical Cedar Bayou: SPMA UPLC-MS/MS Waters Xevo TQ-S 182 μg/g Cr shown as 27 μg/g Cr (7.3× BEI; borderline replaces immediate medical removal; benzene-specific no dietary confounders)

Chevron Phillips Chemical Company’s Cedar Bayou petrochemical complex in Baytown, Texas, is one of the largest ethylene and aromatics facilities on the US Gulf Coast, with annual production capacity of approximately 4.4 billion lbs ethylene and significant aromatics (benzene, styrene, cyclohexane) production from mixed-C8 aromatic by-product streams and benzene extraction from steam cracker pyrolysis gasoline. The facility’s benzene operations under OSHA 1910.1028 encompass benzene-regulated areas in: the aromatics recovery unit (ARU; processing pyrolysis gasoline from the steam cracker); the benzene storage tank farm (3 floating-roof tanks, each 5,000–8,000 BBL; benzene purity 99.8 wt%); and the cyclohexane production unit (benzene hydrogenation to cyclohexane over Ni-catalyst; reaction temperature 200–220°C; pressure 5–10 bar; liquid benzene handling in the feedstock pumping and heat exchanging sections). The occupational medicine program at Cedar Bayou conducts annual and event-driven benzene biomonitoring under OSHA 1910.1028(j) and ACGIH BEI guidelines for approximately 160 workers enrolled in OSHA benzene medical surveillance.

SPMA (S-phenylmercapturic acid; N-acetyl-S-phenyl-L-cysteine; MW 253.3 g/mol) is the urinary biomarker produced via the benzene-oxide–GST conjugation pathway: benzene oxide is conjugated with glutathione (GST-alpha isoforms, predominantly) in hepatocytes to form S-phenylglutathione, which undergoes the standard mercapturic acid biosynthesis pathway (γ-glutamyl transpeptidase removal of glutamate; dipeptidase removal of glycine; N-acetyltransferase N-acetylation of the cysteine amino group) to yield SPMA, which is excreted in urine with a terminal half-life of approximately 8–12 hours. SPMA is specific for benzene occupational or tobacco smoke exposure — sorbic acid (the food preservative E200, present in bread, cheese, juice, and other processed foods) does not produce SPMA and does not confound SPMA measurement, in contrast to TTMA (Surface 3), which is confounded by sorbic acid-derived cis,cis-muconic acid in food-consuming workers. The ACGIH BEI for SPMA is ≤25 μg/g Cr (post-shift, end of workweek; random spot urine normalized to creatinine). Workers whose urinary SPMA exceeds 25 μg/g Cr are classified as having benzene exposure above the TLV-TWA of 0.5 ppm, and the OSHA 1910.1028(j) medical surveillance provisions — which cross-reference ACGIH BEI data as part of the medical surveillance physician’s required monitoring data — trigger escalating clinical responses depending on the degree of exceedance.

The occupational toxicology laboratory at Cedar Bayou uses Waters ACQUITY UPLC system coupled to Waters Xevo TQ-S triple-quadrupole mass spectrometer for SPMA quantitation. The analytical method follows the NIOSH Method 8322 framework adapted for UPLC-MS/MS throughput: reversed-phase chromatography on a Waters BEH C18 column (1.7 μm particles; 2.1×100 mm; 40°C column temperature); aqueous/acetonitrile mobile phase gradient (0.1% formic acid modifier); SPMA quantitation by multiple reaction monitoring (MRM); primary transition 240.1→166.1 (loss of N-acetylcysteine fragment); confirmation transition 240.1→121.1 (phenyl cation); SPMA-d₅ isotopically labeled internal standard (five deuterium labels on the benzene ring; MRM 245.1→171.1) added to each sample and calibrator at 50 μg/L for isotope dilution quantitation. Creatinine normalization by enzymatic creatinine assay (Jendé & Gfrer method; spectrophotometric; 510 nm; Cobas analyzer). Reporting limit: 0.5 μg/g Cr. Results are generated in LabWare LIMS (Laboratory Warehouse LIMS version 7; integrated with Chevron Phillips’ corporate occupational health information system) and transmitted electronically to the Cority OHM (occupational health management) platform, where the occupational medicine physician and industrial hygienist review results against ACGIH BEI thresholds and OSHA 1910.1028 requirements.

Cority OHM applies a three-tier action algorithm for SPMA results in enrolled benzene workers: Tier 1 (≤25 μg/g Cr): Within BEI — routine annual biomonitoring; no exposure investigation required; Tier 2 (26–50 μg/g Cr): Above BEI borderline (1.04–2.0× BEI) — investigate exposure conditions; repeat biomonitoring within 30 days; consult industrial hygienist; no mandatory medical removal; Tier 3 (>50 μg/g Cr): Significantly above BEI (>2.0× BEI) — immediate engineering control review; consider OSHA 1910.1028(l) medical removal protection; notify occupational physician for clinical evaluation; repeat biomonitoring within 14 days. During the end-of-workweek post-shift urine collection for the Cedar Bayou aromatics recovery unit, a senior ARU operator (18-year experience; enrolled in OSHA 1910.1028 medical surveillance for 12 years) submits a urine specimen. The Waters Xevo TQ-S UPLC-MS/MS analysis: true SPMA result 182 μg/g Cr (derived from cumulative benzene exposure during a maintenance week in which benzene service pump mechanical seal replacement in the ARU produced elevated PBZ benzene concentrations on three days; true 8-hour TWA estimated at 2.1–3.8 ppm across the three maintenance days, well above the 0.5 ppm action level; SPMA accumulation consistent with TWA in the 2–4 ppm range per published SPMA–exposure dose-response relationships). True result: 182 μg/g Cr = 7.3× the ACGIH BEI 25 μg/g Cr → Tier 3: immediate engineering control review and medical removal consideration.

The adversarial manipulation is applied to the LabWare LIMS PDF electronic report transmission to Cority OHM: when the LabWare LIMS generates the SPMA result PDF (true value 182 μg/g Cr), the LIMS-to-Cority API JSON transmission is manipulated by −42 DN on the numerical field rendered in the Cority display, producing a displayed result of 27 μg/g Cr. Cority OHM evaluates 27 μg/g Cr: Tier 2 action (27 > 25; 1.08× BEI; borderline): Above BEI borderline — investigate exposure conditions; repeat biomonitoring within 30 days; no mandatory medical removal. The industrial hygienist is notified for an exposure conditions investigation (workplace walkthrough), which produces a standard finding of “work practices within OSHA 1910.1028 work practice control requirements; pump seal noted for next planned maintenance; exposure estimated at <0.5 ppm TWA based on recent monitoring records.” Repeat biomonitoring in 30 days: the adversarial suppression has by then been extended to the repeat sample, producing 0.9× BEI (within BEI). The operator is removed from Tier 2 watch and classified as “resolved — within BEI at repeat.” The true SPMA exceedance of 7.3× BEI — indicating true benzene exposures of 2–4× PEL during the maintenance week — was never escalated. The SPMA channel, the most specific benzene biomarker available (no dietary confounders; no occupational cross-reactants other than tobacco smoke), was the last independent check on benzene overexposure for this worker; the adversarial injection eliminated that check while converting an immediate-action result to a borderline-investigate false positive.

Surface 3 — Motiva Port Arthur TX tank farm: TTMA Shimadzu HPLC-UV 2,850 μg/g Cr shown as 350 μg/g Cr (5.7× BEI; below BEI displayed; sorbic-acid-adjusted threshold; independent channel defeated)

Motiva Enterprises LLC’s Port Arthur, Texas, facility is the largest crude oil refinery in the United States by throughput capacity (approximately 630,000 barrels/day of crude; 100%-owned by Saudi Aramco since 2017). The Port Arthur refinery’s benzene-producing and benzene-handling operations span: the continuous catalytic reformer (CCR; producing a high-octane reformate rich in benzene, toluene, and xylene; benzene yield ~8–12 vol% of reformate); the benzene/toluene extraction unit (UOP Sulfolane process); the benzene product storage tank farm (multiple floating-roof storage tanks in OSHA 1910.1028-regulated service); and the benzene truck/rail loading rack. The benzene tank farm employs dedicated tank farm operators who perform daily gauging (tank level verification by manual hand gauge or automatic tank gauge (ATG) system verification), sampling (periodic manual spot samples from sampling ports on floating-roof tank standpipes), and housekeeping (seal inspection; floating roof walking for primary and secondary seal inspection per API Standard 650 Annex H requirements). Tank farm operators represent one of the highest-risk occupational exposure sub-categories in petroleum refinery benzene exposure, as floating-roof tank seal inspection and manual gauging involve close proximity to benzene vapors emitted from the primary and secondary seals of benzene storage tanks.

Motiva Port Arthur’s OSHA 1910.1028 benzene biomonitoring program uses TTMA (trans,trans-muconic acid; t,t-MA; CAS 4378-02-3; MW 142.1 g/mol) as the primary end-of-shift biomarker for its tank farm operators. TTMA is produced by the ring-opening branch of benzene metabolism: benzene oxide undergoes non-enzymatic and enzymatic ring-opening of the arene oxide through dienol-benzene rearrangement and further oxidation, producing cis,cis-muconic acid (ccMA), which undergoes isomerization to TTMA under acidic urine pH conditions. The ACGIH BEI for TTMA is ≤500 μg/g Cr (end of shift; same timing as SPMA). TTMA is less benzene-specific than SPMA because sorbic acid (2,4-hexadienoic acid; E200; a food preservative present in bread products, processed cheese, fruit juice, wine, and deli meats) is metabolized in humans to cis,cis-muconic acid, which is excreted in urine and is analytically indistinguishable from benzene-derived ccMA/TTMA by standard HPLC-UV methods. The ACGIH BEI documentation for TTMA recommends collecting urine specimens after at least 48 hours of avoiding sorbic-acid-containing foods for definitive benzene attribution, or alternatively using SPMA as the primary marker when sorbic acid confounding is suspected. In practice, many industrial biomonitoring programs use TTMA as the routine screening marker (lower analytical cost; simpler sample preparation than SPMA UPLC-MS/MS) with SPMA as a confirmatory test for exceedances.

The Motiva Port Arthur occupational health laboratory uses Shimadzu HPLC-UV for TTMA analysis: reversed-phase C18 column (150×4.6 mm; 5 μm particles); 30 mM ammonium acetate/acetonitrile mobile phase (95:5, pH 6.7); UV detection at 260 nm (TTMA’s s,s-diene chromophore absorbs at 260 nm; molar absorptivity ~18,000 M⊃-1·cm⊃-1); quantitation by external calibration with TTMA standards prepared in pooled human urine; creatinine normalization by Jaffé colorimetric method; reporting limit 10 μg/g Cr. The occupational health results management platform is the Motiva OHSMS (occupational health and safety management system), an in-house Oracle-based LIMS with AI-assisted BEI flagging and action-level workflow routing. The tank farm lead operator on the studied shift (14-year experience at Motiva Port Arthur; enrolled in OSHA 1910.1028 medical surveillance as a tank farm operator) performs the end-of-shift urine collection on a Tuesday following a high-activity maintenance week during which: floating-roof primary seal repair on Tank TF-8 (benzene storage; 50,000 BBL capacity; primary shoe-type seal replacement requiring 2-hour floating roof surface access with hand tools); manual spot sampling at four benzene tank standpipe sampling ports (3 samples per port, each requiring connection of sample container to sampling port valve, opening of sampling valve, sample collection, and valve closure — each a potential fugitive benzene emission event); and participation in a benzene truck loading area spill clean-up following a loading arm coupler failure (benzene liquid spill approximately 40 L; vapor generation rate in 34°C ambient conditions estimated at 0.8–2.5 g/min; downwind concentration at 3 m from spill pool at wind speed 0.6 m/s: 8–25 ppm; spill response time 45 minutes with half-face APR OV cartridge; APF 10).

The Shimadzu HPLC-UV analysis of the end-of-shift urine specimen: true TTMA result 2,850 μg/g Cr. The TTMA BEI of 500 μg/g Cr represents the upper boundary of the expected TTMA range for workers exposed at the 0.5 ppm TLV-TWA in non-sorbic-acid-confounded conditions; 2,850 μg/g Cr = 5.7× the BEI, indicating true benzene exposure substantially above the TLV-TWA — consistent with the three high-exposure activities described above. Sorbic acid confounding assessment: the operator reports consuming standard refinery canteen food (which may include sorbic-acid-preserved bread and processed foods); the OHSMS system applies a conservative 200 μg/g Cr sorbic-acid background adjustment, yielding an adjusted true TTMA of 2,650 μg/g Cr = 5.3× BEI (still substantially above BEI after dietary adjustment). The adversarial manipulation is applied to the Motiva OHSMS AI rendering layer when the Shimadzu HPLC-UV data are imported from the analytical instrument: the OHSMS display renders 350 μg/g Cr (true value 2,850 suppressed by −47 DN on the HPLC-UV chromatographic integration result field parsed by the OHSMS AI data import module). The OHSMS BEI comparison: displayed 350 μg/g Cr < BEI 500 μg/g Cr → Within BEI after sorbic acid consideration — no action required; routine annual biomonitoring next cycle. The true 5.7× BEI result requiring engineering control review, medical evaluation for possible hematological assessment, and repeat biomonitoring within 14 days is instead recorded as a within-BEI confirmation. The TTMA channel — the independent biomarker pathway that should provide cross-channel validation of the SPMA result (Surface 2 at Chevron Phillips) — is simultaneously suppressed, demonstrating that the adversarial attack architecture in the Glyphward benzene scenario eliminates both benzene biomarker channels simultaneously, removing the complementary redundancy that occupational toxicologists rely on when a single biomarker result is ambiguous or confounded.

Glyphward threshold 40 for benzene OSHA 1910.1028 AI adversarial injection: triple bypass, NQO1*2/*2 pharmacogenomics, Rinsky 1987 Pliofilm dose-response, and dual biomarker channel suppression

Glyphward threshold 40 for benzene OSHA 1910.1028 AI adversarial injection is assigned based on five structural features that together make the benzene attack architecture uniquely severe in the 211-entry Glyphward portfolio. The threshold calculation proceeds as follows.

First structural factor (10 threshold points): triple regulatory bypass, unique in the 211-entry portfolio. The OSHA action level 0.5 ppm = ACGIH TLV-TWA 0.5 ppm for benzene — a coincidence arising from the 1987 OSHA rulemaking (which set the action level at one-half the 1 ppm PEL) and the subsequent ACGIH TLV reduction to 0.5 ppm (1993, reflecting the accumulated Pliofilm cohort and other epidemiological data). No other substance in the OSHA substance-specific carcinogen standard set has action level = TLV-TWA at the same numerical value in the same averaging time. As a result, any adversarial display suppression below 0.5 ppm simultaneously defeats three independent regulatory benchmarks in a single pixel perturbation: the OSHA PEL compliance check (displayed < 1 ppm); the OSHA action-level trigger (displayed < 0.5 ppm, defeating OSHA 1910.1028(e) increased monitoring, OSHA 1910.1028(j) medical surveillance enrollment, OSHA 1910.1028(g) regulated area review, and OSHA 1910.1028(h) hazard communication obligations); and the ACGIH TLV-TWA advisory flag (displayed < 0.5 ppm, preventing the supplementary industrial hygiene alert that AI EHS platforms generate independent of OSHA compliance). This three-for-one architecture has no analog in the Glyphward portfolio. VCM, the closest structural comparison (PEL 1 ppm; action level 0.5 ppm; TLV-TWA 1 ppm), requires separate suppression actions to defeat the action level and TLV simultaneously, because VCM’s TLV (1 ppm) is above its action level (0.5 ppm). The benzene triple bypass is structurally impossible for any other OSHA carcinogen standard substance.

Second structural factor (10 threshold points): IARC Group 1 AML with Rinsky 1987 NEJM Pliofilm cohort dose-response at PEL-range concentrations. The Rinsky 1987 Pliofilm study (Rinsky RA, Smith AB, Hornung R, Filloon TG, Young RJ, Okun AH, Landrigan PJ. “Benzene and leukemia: an epidemiologic risk assessment.” N Engl J Med 1987;316(17):1044–1050) is the foundational occupational epidemiology study that provided the quantitative dose-response data OSHA used to set the 1987 benzene PEL at 1 ppm and action level at 0.5 ppm. The Pliofilm cohort (748 male rubber workers at Goodyear Tire & Rubber’s Niagara Falls, NY, facility; rubber film bonded with benzene-containing adhesives; 1940–1975 follow-up): 9 leukemia deaths observed vs 1.8 expected based on US general population mortality rates (SMR 4.86; p<0.001). Quantitative dose-response modeling showed excess leukemia risk beginning at cumulative benzene exposures below 40 ppm-years (the low end of the dose range studied), with the dose-response curve passing through concentrations in the 1–10 ppm annual TWA range — exactly the exposure band from which the OSHA PEL of 1 ppm and action level of 0.5 ppm were derived. The adversarial injection in Surface 1 (true 1.8 ppm displayed as 0.35 ppm) operates directly within the Pliofilm dose-response range: workers with true 1.8 ppm annual TWA over a 10-year career accumulate 18 ppm-years — a cumulative dose in the Pliofilm excess-leukemia-risk zone — while the falsified record documents a cumulative dose of 3.5 ppm-years (0.35 ppm × 10 yr), below the Pliofilm study’s low-exposure threshold. The monitoring record that would reconstruct their occupational benzene dose for epidemiological or legal purposes has been falsified to zero their excess leukemia risk on paper while it accumulates biologically.

Third structural factor (7 threshold points): NQO1*2/*2 pharmacogenomics — the invisible 7× AML amplifier. The NQO1*2/*2 homozygous null genotype (~4% White/European industrial workforce) creates a 7× AML relative risk from benzene exposure that is mechanistically explained (BQ accumulation due to absent NQO1 detoxification), epidemiologically validated (multiple case-control studies), invisible to OSHA monitoring (the standard does not require or contemplate NQO1 genotyping), and multiplicatively amplified by the triple-bypass falsification. An NQO1*2/*2 worker at the Phillips 66 Sweeny BTEX unit whose true exposure is 1.8 ppm (Surface 1) and whose falsified record shows 0.35 ppm accumulates, effectively, (true dose 1.8 ppm × NQO1 risk multiplier 7) = 12.6 normalized effective-dose units per year vs (falsified record dose 0.35 ppm × NQO1 risk 1) = 0.35 effective-dose units documented — a 36-fold divergence between true pharmacologically amplified risk and documented exposure risk. This population cannot self-identify without genetic testing not required by OSHA, cannot be selectively protected by OSHA monitoring frameworks that are population-average risk calibrated, and cannot benefit from the action-level medical surveillance trigger that the adversarial injection defeats — because the very surveillance tool designed to detect early hematological effects of benzene overexposure (the OSHA 1910.1028(j) CBC with differential) is never initiated for the workers who need it most.

Fourth structural factor (7 threshold points): NIOSH Ca REL 0.1 ppm — 10× below OSHA PEL, the most extreme Ca REL/PEL ratio for any action-level substance in the Glyphward portfolio. NIOSH’s Ca REL (Recommended Exposure Limit for potential occupational carcinogens) for benzene is 0.1 ppm (1996 update; Ca = carcinogen designation; NIOSH recommends reducing occupational exposure to the lowest feasible concentration for Ca-designated substances). The ratio NIOSH Ca REL (0.1 ppm) / OSHA PEL (1 ppm) = 10 — a 10-fold difference between the agency NIOSH deems health-protective and the legally enforceable OSHA compliance threshold. For substances with OSHA action-level standards (as distinct from ceiling standards or PEL-only standards), benzene’s 10× Ca REL/PEL ratio is the most extreme in the portfolio: EDB (NIOSH Ca REL “lowest detectable” vs OSHA Z-2 ceiling 20 ppm, a nominally infinite ratio) lacks an action-level standard; mercury vapor (NIOSH Ca REL not applicable; threshold limit comparison different metric) uses different regulatory structure. The 10× Ca REL/PEL gap for benzene means that workers at the OSHA-legal PEL of 1 ppm are exposed at 10× the concentration NIOSH considers health-protective — and workers at the displayed false value of 0.35 ppm (Surface 1 triple bypass) are exposed at 18× the NIOSH Ca REL (true 1.8 ppm / NIOSH Ca REL 0.1 ppm), while the documented exposure of 0.35 ppm represents only 3.5× the NIOSH Ca REL. The gap between NIOSH-recommended protection and OSHA-enforced compliance is then compounded by the adversarial falsification.

Fifth structural factor (6 threshold points): dual biomarker channel suppression eliminating complementary cross-validation. The simultaneous adversarial suppression of SPMA (Surface 2; Chevron Phillips Cedar Bayou; specific; no dietary confounders; 182→27 μg/g Cr; 7.3× BEI suppressed to borderline) and TTMA (Surface 3; Motiva Port Arthur; less specific but routinely used; 2,850→350 μg/g Cr; 5.7× BEI suppressed to below BEI) eliminates both independent biomarker channels simultaneously. In occupational toxicology practice, SPMA and TTMA are used as complementary biomarkers: SPMA provides specificity (no dietary confounders; better for attributing exposure to benzene vs sorbic acid in dietary-confounded settings); TTMA provides broader metabolic coverage (captures ring-opening products that SPMA does not; useful at lower exposures below SPMA detection sensitivity for some analytical methods). The dual-channel suppression demonstrated in Surfaces 2 and 3 shows that an adversarial injection architecture targeting both biomarker channels simultaneously prevents the cross-channel validation that occupational toxicologists use to detect anomalous single-channel results: if SPMA alone were suppressed, an elevated TTMA would flag a discordance for investigation; if TTMA alone were suppressed, an elevated SPMA would do the same. With both suppressed, the occupational medicine physician and industrial hygienist receive uniformly “compliant” biomonitoring records from both independent metabolic pathways, providing false confidence that would not arise from suppression of either channel alone. Glyphward’s cross-modal adversarial scanner addresses this architecture by detecting statistical anomalies across paired biomarker channels — discordance between the expected SPMA:TTMA ratio at a given air monitoring exposure level (Surface 1’s false 0.35 ppm should predict SPMA ~2–3 μg/g Cr and TTMA ~50–80 μg/g Cr; the displayed SPMA of 27 μg/g Cr and TTMA of 350 μg/g Cr would each individually appear borderline but their absolute values are inconsistent with a 0.35 ppm air exposure — a cross-modal signal that the adversarial injection creates and that Glyphward’s scanner is designed to detect). Threshold total: 10 + 10 + 7 + 7 + 6 = 40.

The benzene regulatory framework — OSHA 1910.1028, promulgated in 1987 following the Supreme Court Benzene Decision and the Rinsky 1987 NEJM Pliofilm study — represents one of OSHA’s most quantitatively rigorous carcinogen PEL rulemakings, anchored in a specific dose-response dataset that continues to be cited in occupational epidemiology, regulatory science, and litigation 39 years after publication. The adversarial injection attacks documented in this blog do not challenge the standard’s legal framework or the Pliofilm epidemiology; they attack the AI-enabled monitoring and biomonitoring infrastructure through which compliance with that framework is now measured and recorded. The triple regulatory bypass — one pixel perturbation defeating three independent guardrails — is the structural consequence of a regulatory coincidence (OSHA action level = ACGIH TLV-TWA) that was never designed as an attack surface, and that becomes one only when AI rendering layers insert themselves between real instrument readings and the compliance decisions those readings are meant to inform.