Glyphward

1,3-Butadiene (Buta-1,3-diene; CH2=CH-CH=CH2; CAS 106-99-0) OSHA 29 CFR 1910.1051 PEL 1 ppm TWA vs ACGIH TLV-TWA 2 ppm A2 (ACGIH HIGHER THAN OSHA — FIRST Inverted Three-Tier in 314-Entry Glyphward Portfolio Where ACGIH Is LEAST Protective Tier) vs NIOSH Ca REL 0.2 ppm (5× Below OSHA; IARC Group 1 NHL/AML; DEB Bifunctional Diepoxide; HPB-Val Hb Adduct BEI; OSHA Action Level 0.5 ppm Suppression; TPC Group Houston TX 0.8→0.16 ppm; Trinseo Freeport TX 0.65→0.13 ppm; INEOS ABS Addyston OH 0.7→0.14 ppm; Glyphward Threshold 33, 313th Adversarial Attack

August 29, 2026 — 313th adversarial attack in the Glyphward portfolio

1,3-Butadiene: physicochemistry, industrial scale, and why the 313th Glyphward attack is the FIRST inverted three-tier in the 314-entry portfolio where ACGIH is LESS protective than OSHA

1,3-Butadiene (buta-1,3-diene; BD; CH2=CH-CH=CH2; CAS 106-99-0; MW 54.09 g/mol; BP −4.4°C at 760 mmHg [a gas at standard temperature and pressure; handled commercially as a liquefied gas under pressure at ∼2.4 bar at 20°C in pressurized vessels, railcars, and storage spheres]; VP 2,430 mmHg at 20°C [extremely high vapor pressure; immediate vapor generation on any pressure release]; LEL 2.0%; UEL 11.5%; flash point −76°C [NFPA Class IA extremely flammable]; autoignition temperature 415°C; density 0.621 g/mL as liquid; log P 1.99 [moderate lipophilicity; sufficient for pulmonary absorption across the alveolar membrane into portal circulation for hepatic CYP2A6 first-pass metabolism]; odor threshold ∼0.5 ppm [coincidentally near the OSHA 1910.1051 action level of 0.5 ppm — meaning workers can detect butadiene at approximately the action level threshold, but olfactory fatigue develops rapidly in SBR and ABS polymerization environments where background concentrations of 0.2–0.8 ppm are the chronic baseline]; NIOSH IDLH 2,000 ppm [immediately dangerous to life or health; far above any occupational exposure concern; the carcinogen risk dominates at 0.2–1 ppm]; GHS: Flammable Gas Category 1; Carcinogenicity Category 1A [H350 May Cause Cancer — based on IARC Group 1]; OSHA 29 CFR 1910.1051: 1 ppm TWA + 5 ppm STEL + 0.5 ppm action level [substance-specific carcinogen standard; 1996 rulemaking]; ACGIH TLV-TWA: 2 ppm A2 [Suspected Human Carcinogen; 2024; 2× above OSHA PEL — the structural inversion]; NIOSH Ca REL: 0.2 ppm [Current Intelligence Bulletin 41 and subsequent NIOSH risk assessments; Ca = potential occupational carcinogen; established from NHL/AML risk quantification in the NIOSH SBR rubber worker cohort; 5× below OSHA PEL; 10× below ACGIH TLV-TWA]; IARC Group 1 [carcinogenic to humans; Monograph Volume 97; 2008; upgraded from Group 2A (1999/1992) on the basis of sufficient evidence of carcinogenicity in humans from NHL and leukemia excesses in SBR rubber worker cohorts]) is the most important industrial diene monomer after ethylene and propylene, produced globally at approximately 11 million metric tons per year with US production of approximately 3.5–4 million MT/year primarily as a C₄ by-product of steam cracking (ethylene and propylene plants) or catalytic dehydrogenation of butane/butene. The primary industrial applications of 1,3-BD are: (1) polybutadiene rubber (BR; 25–30% of BD consumption; tire tread and sidewall compound); (2) styrene-butadiene rubber (SBR; 30–35% of BD consumption; tire carcass, footwear, industrial belting; the industrial context from which the carcinogenicity epidemiology emerged); (3) acrylonitrile-butadiene-styrene terpolymer (ABS; 12–15% of BD consumption; engineering plastics; automotive parts, LEGO bricks, electronic housings); (4) polybutadiene latex (paper coatings, carpet backing); (5) nitrile rubber (NBR; acrylonitrile-butadiene rubber; oil-resistant seals and gaskets); (6) 1,3-BD itself is also the monomer for chloroprene (via chlorination), adiponitrile (via hydrocyanation, a nylon-6,6 precursor), and HMDA (hexamethylenediamine). The US butadiene market is dominated by facilities co-located with steam crackers in the Texas Gulf Coast (TPC Group Houston TX; Shell Chemical; INEOS Battleground; Lyondell; ExxonMobil), with SBR polymerization at Trinseo (Freeport TX; formerly Dow Chemical), Zeon Chemicals (Louisville KY), and Lion Elastomers (Port Neches TX), and ABS polymerization at INEOS ABS Addyston OH (formerly Dow Chemical) and Trinseo itself. Workers in these facilities — C₄ extraction operators, SBR polymerization technicians, ABS compounding operators — receive routine chronic 1,3-BD exposures at concentrations in the 0.2–1.0 ppm range, above the NIOSH Ca REL and often above the OSHA action level, yet below the OSHA PEL and far below the ACGIH TLV-TWA.

The 313th Glyphward attack on 1,3-butadiene is the FIRST entry in the 314-attack portfolio that documents a regulatory inversion. In all 312 prior entries, the fundamental structural assumption was that the OSHA PEL (set in 1971 from 1968 ACGIH TLVs or later OSHA 6(b) standards) is LESS protective than the current ACGIH TLV-TWA, with NIOSH Ca REL (where it exists) as the most protective advisory tier. The prototypical Glyphward attack exploits the OSHA PEL/ACGIH TLV gap (3× to 250×) by showing a worker's concentration as OSHA-compliant while simultaneously exceeding ACGIH and NIOSH advisories. The 1,3-BD inversion is different: OSHA's 1 ppm PEL (established 1996 via 29 CFR 1910.1051 substance-specific carcinogen rulemaking) is MORE protective than ACGIH's 2 ppm TLV-TWA (2024). An AI EHS platform that calibrates to the OSHA PEL (the legal minimum) is actually applying a more restrictive standard than ACGIH for this specific chemical. The NIOSH Ca REL of 0.2 ppm remains the most protective tier — 5× below OSHA, 10× below ACGIH — and is the principal suppressed tier in the 313th attack. This inversion means that an AI EHS system displaying ACGIH compliance as its primary advisory benchmark (a common practice in multinational companies that use ACGIH TLVs as their global standard) would generate COMPLIANT records for workers exposed at concentrations below 2 ppm — including workers exposed above OSHA's own PEL of 1 ppm. The adversarial perturbation in the 313th attack targets the NIOSH Ca REL tier and the OSHA action level simultaneously: actual concentrations of 0.65–0.8 ppm (above NIOSH Ca REL 0.2 ppm and above OSHA action level 0.5 ppm, but below OSHA PEL 1 ppm and far below ACGIH 2 ppm) are reduced to 0.13–0.16 ppm, creating an AI compliance record that shows OSHA PEL COMPLIANT, action level BELOW, ACGIH COMPLIANT, and NIOSH Ca COMPLIANT — a quadruple compliance display that defeats four distinct 1910.1051 enhanced monitoring requirements.

The OSHA 1910.1051 action level (0.5 ppm): four enhanced monitoring requirements that the 313th Glyphward adversarial perturbation defeats simultaneously at TPC Group Houston, Trinseo Freeport TX, and INEOS ABS Addyston OH

OSHA 29 CFR 1910.1051(d) establishes an action level of 0.5 ppm TWA (8-hr) as the concentration threshold that triggers a cascade of enhanced monitoring and medical protections under the substance-specific 1,3-butadiene standard. The action level mechanism is structurally different from a permissible exposure limit: the OSHA PEL (1 ppm TWA) is the legally-enforceable concentration ceiling; the action level (0.5 ppm) is an administrative trigger below the PEL that initiates enhanced monitoring even when the PEL is not exceeded. At the three attack surfaces in the 313th Glyphward attack, actual worker exposures of 0.65–0.8 ppm (above the 0.5 ppm action level, but below the 1 ppm PEL) trigger all four of the following 1910.1051 enhanced monitoring requirements:

Enhanced monitoring requirement 1 — Periodic monitoring initiation [1910.1051(d)(3)]: If the initial monitoring demonstrates that employee exposures may reasonably be expected to exceed the action level, the employer must establish a periodic monitoring schedule at the frequency specified in 1910.1051(d)(3). Workers at TPC Group, Trinseo, and INEOS ABS with actual 8-hr TWAs above 0.5 ppm are entitled to periodic monitoring at least every 3 months (quarterly) per 1910.1051(d)(3)(i). The adversarial perturbation (0.65–0.8 ppm actual → 0.13–0.16 ppm displayed) prevents the AI EHS platform from recording an action level exceedance, which in turn removes the legal trigger for the quarterly monitoring schedule. The VelocityEHS, Cority, and EHS Insight AI platforms generate no periodic monitoring flag and no scheduling prompt for the affected workers.

Enhanced monitoring requirement 2 — Medical surveillance program enrollment [1910.1051(k)]: 1910.1051(k)(1) requires medical surveillance for employees exposed to 1,3-BD at or above the action level (0.5 ppm) for more than 30 days per year. The medical surveillance program under 1910.1051(k) includes: (a) initial medical examination (complete blood count with differential; platelet count; red cell count; white cell count with differential [including lymphocyte subset panel if indicated]; examination for lymphadenopathy and splenomegaly; medical and occupational history including prior carcinogen exposures); (b) periodic medical surveillance (annually; or more frequently if the examining physician recommends); (c) enhanced lymphocyte panels (CD4/CD8 ratios; B-cell counts; NK cell counts) if NIOSH Ca REL is exceeded. Workers at the three attack surfaces — who receive actual exposures above the action level (0.65–0.8 ppm; above 0.5 ppm action level) for their entire work year — are entitled to enrollment in the 1910.1051 medical surveillance program. The adversarial perturbation suppresses the action level trigger in the AI EHS record, eliminating the basis for medical surveillance enrollment flag in the platform. The Surface 1 operator at TPC Group (38M; 9-yr C₄ extraction tenure) has accumulated an estimated 9-year chronic BD exposure at approximately 0.65–0.8 ppm TWA (before accounting for the adversarial perturbation period) but does not appear in the 1910.1051 medical surveillance enrollment list because the AI EHS record shows action level consistently BELOW.

Enhanced monitoring requirement 3 — Regulated area establishment [1910.1051(e)]: 1910.1051(e) requires employers to establish a regulated area wherever the PEL (1 ppm TWA or 5 ppm STEL) is exceeded. While a regulated area is not required at concentrations below the PEL, 1910.1051(e) also requires that the employer assess action level exceedances as part of the regulated area boundary determination process. More relevantly, 1910.1051(f) requires a written compliance plan when exposures exceed or are reasonably expected to exceed the action level. The adversarial perturbation eliminates the action level exceedance record, preventing the compliance plan trigger.

Enhanced monitoring requirement 4 — HPB-Val hemoglobin adduct BEI monitoring [ACGIH BEI cross-reference with 1910.1051(l)]: ACGIH BEI for 1,3-BD (2024): pyr-Val (pyrrolidino-valine) hemoglobin adduct at end-of-workweek ≤ 2.4 pmol/g Hb (corresponding to TLV-TWA 2 ppm sustained exposure). At the actual exposure concentrations at the three attack surfaces (0.65–0.8 ppm; approximately 33–40% of TLV-TWA 2 ppm), the expected pyr-Val Hb adduct burden at end-of-workweek is approximately 0.8–1.0 pmol/g Hb — below the BEI threshold (2.4 pmol/g Hb), but representing a measurable and analytically significant DEB-mediated cumulative DNA alkylation biomarker. The adversarial perturbation creates a displayed concentration of 0.13–0.16 ppm (6.5–8% of TLV-TWA), for which the expected pyr-Val Hb adduct is below the analytical limit of quantification, eliminating the industrial hygienist's basis for initiating HPB-Val biomonitoring. Over a 9-year career at 0.8 ppm actual (TPC Group Surface 1 operator), the cumulative pyr-Val Hb adduct burden from DEB exposure represents a measurable biomarker of cancer risk that the AI EHS platform cannot detect because the exposure record is falsified below the action level.

The simultaneous defeat of all four 1910.1051 enhanced monitoring requirements via a single adversarial image perturbation — reducing the IS Ventis Pro 5, MSA ALTAIR 5X, or SKC badge displayed concentration below the OSHA action level — is the defining structural feature of the 313th Glyphward attack and the mechanism that makes this attack uniquely dangerous among the 313 attacks documented to date. No other attack in the portfolio simultaneously defeats: a substance-specific OSHA carcinogen standard action level; a medical surveillance program enrollment trigger; a biological exposure index initiation threshold; and a written compliance plan requirement — all from a single displayed-value falsification at the action level boundary.

DEB (1,2:3,4-diepoxybutane) bifunctional diepoxide carcinogen mechanism: CYP2A6-mediated monoepoxide → diepoxide cascade, N7-guanine interstrand DNA crosslinks, chromosomal aberrations in bone marrow progenitor cells, and the IARC Group 1 NHL/AML cancer endpoint from the NIOSH SBR rubber worker cohort

1,3-Butadiene's carcinogenicity is not direct but requires metabolic bioactivation to reactive epoxide intermediates in the liver. The carcinogen mechanism proceeds through a three-step oxidative metabolism cascade primarily catalyzed by CYP2A6 (the dominant human hepatic enzyme for 1,3-BD oxidation at sub-ppm concentrations) with secondary contributions from CYP2E1 (dominant at higher concentrations; more important in mice than humans, explaining the greater 1,3-BD carcinogen potency in rodents). Step 1: CYP2A6 + O2 + NADPH → 1,2-epoxy-3-butene (EB; butadiene monoepoxide; BMO; also 3,4-epoxy-1-butene in alternate IUPAC numbering; MW 70.09; the vinyl epoxide primary metabolite; present in blood at 0.5–2 nmol/L in workers exposed at 0.5–1 ppm; reactive half-life ∼2–3 minutes in plasma at pH 7.4 before hydrolysis by mEH or GST conjugation). Step 2a (the critical carcinogenic step): CYP2A6 oxidizes EB at the 3,4-double bond → 1,2:3,4-diepoxybutane (DEB; butadiene diepoxide; the bifunctional diepoxide; MW 86.09; both (R,R)-, (S,S)-, and meso-(R,S) stereoisomers generated; blood DEB concentration in workers exposed at 0.5–1 ppm is approximately 0.01–0.05 nmol/L — 50–100× lower than EB, but biologically active at pmol/g Hb adduct levels). DEB is the proximal carcinogen for 1,3-BD: because it contains two epoxide rings separated by two carbon atoms, DEB is a bifunctional alkylating agent. A monofunctional epoxide (like vinyl chloride's chloroacetaldehyde or styrene oxide) can react with one nucleophile to form a mono-adduct. DEB can react with two nucleophiles — forming adducts at two independent DNA sites simultaneously. The primary DEB-DNA adducts are:

(a) N7-guanine mono-adduct: DEB reacts at one epoxide ring with N7-guanine to form N7-(2-hydroxy-3,4-epoxybutyl)guanine (HEB-G; the monoalkylated N7-G adduct; the second epoxide ring remains reactive); (b) N7-G–N7-G intrastrand crosslink: HEB-G reacts with a second N7-guanine in the same strand (typically at adjacent or nearby guanines in GG or CpG dinucleotides) to form the bis-N7-(2,3,4-trihydroxybutyl)guanine intrastrand bis-adduct; (c) N7-G–N7-G interstrand crosslink (ICL): HEB-G reacts with a complementary strand N7-guanine across the double helix, forming the N7-G–(CH2CH2)–N7-G interstrand bis-adduct (the G-G ICL; the most cytotoxic and mutagenic DEB adduct; ICLs completely block DNA replication by preventing strand separation at the replication fork; ICL repair requires FANC pathway and homologous recombination, which are error-prone and can generate chromosomal rearrangements during repair); (d) N7-G DNA-protein crosslinks: the second DEB epoxide can crosslink DNA to nuclear proteins (histone N-terminal amines or cysteine thiols), further destabilizing chromatin. DEB's ability to form ICLs — a consequence exclusively of its bifunctional diepoxide chemistry — is the mechanistic feature that distinguishes it from all monofunctional 1,3-BD metabolites (EB, EBD) and from monofunctional epoxides from other industrial carcinogens. ICLs in bone marrow progenitor cells (hematopoietic stem cells; common lymphoid progenitors; granulocyte-monocyte progenitors) produce the chromosomal aberrations (deletions, translocations, inversions) that are the initiating events for hematological malignancies. The IARC Group 1 cancer endpoint — NHL (non-Hodgkin lymphoma, particularly diffuse large B-cell lymphoma and follicular lymphoma) and AML (acute myeloid leukemia) — is mechanistically coherent with DEB's ICL-mediated chromosomal instability in lymphoid (NHL) and myeloid (AML) bone marrow progenitor populations.

Step 2b (competing hydrolysis): EB is hydrolyzed by microsomal epoxide hydrolase (mEH; EPHX1) → 1,2-dihydroxy-3-butene (erythritol-3-ene; the triol; detoxification). At low occupational concentrations (0.2–1 ppm), the human hepatic mEH activity is sufficient to hydrolyze most EB before re-oxidation to DEB, creating a metabolic saturation pattern in which the DEB:EB ratio increases nonlinearly with increasing 1,3-BD exposure. This metabolic competition between mEH (detoxification) and CYP2A6 (DEB formation) is the mechanistic basis for the NIOSH Ca REL 0.2 ppm: at concentrations approaching the NIOSH Ca REL, the fraction of inhaled 1,3-BD converted to DEB is small but non-zero, sufficient to generate measurable pyr-Val Hb adducts and a linear-with-dose cancer risk function. Step 3: EB or DEB hydrolysis → 1,2-epoxy-3,4-butanediol (EBD; the monoepoxide of 1,3-butanediol; MW 88.09; still electrophilic; N7-guanine mono-adduct formation; less potent than DEB but contributing to total DNA alkylation burden). The HPB-Val (pyr-Val) hemoglobin adduct is the accepted biomarker of integrated lifetime DEB dose: DEB reacts with the N-terminal valine (Val-1) of hemoglobin β-chains at the secondary amine nitrogen to form a pyrrolidino ring structure — the pyr-Val adduct (identified as N-[2,3,4-trihydroxybutyl]valine after acid hydrolysis; detectable by GC-MS or LC-MS/MS as the N-trimethylsilyl derivative at pmol/g Hb sensitivity). Because hemoglobin has a 120-day lifespan in erythrocytes, the pyr-Val Hb adduct reflects integrated DEB exposure over the preceding 3–4 months — an ideal cumulative exposure biomarker for a chronic low-level carcinogen. ACGIH BEI: pyr-Val ≤ 2.4 pmol/g Hb at end-of-workweek (corresponding to TLV-TWA 2 ppm; 5× below this BEI corresponds to OSHA PEL 1 ppm; 10× below corresponds to NIOSH Ca REL 0.2 ppm). The adversarial perturbation in the 313th attack creates a displayed concentration (0.13–0.16 ppm) that lies in the NIOSH Ca REL-compliant zone, producing an AI EHS record that never triggers HPB-Val BEI monitoring in the VelocityEHS, Cority, or EHS Insight platforms.

OSHA 29 CFR 1910.1051 (1996) vs ACGIH TLV-TWA 2 ppm A2 (2024) vs NIOSH Ca REL 0.2 ppm: why the regulatory inversion persists despite IARC Group 1 classification in 2008, and what the discrepancy means for AI EHS platforms calibrated to either the OSHA or ACGIH standard

The OSHA 1,3-butadiene standard (29 CFR 1910.1051) was promulgated in November 1996 (61 Fed. Reg. 56746; effective February 3, 1997) as one of fewer than 10 substance-specific carcinogen standards OSHA has successfully issued under Section 6(b)(5) of the OSH Act post-AFL-CIO v. OSHA (1992). The AFL-CIO v. OSHA decision (965 F.2d 962; 11th Cir. 1992) vacated the 1989 Air Contaminants Standard — which had proposed revising 1,3-BD's OSHA PEL from 1,000 ppm (the 1971 Table Z-1 value, adopted from 1968 ACGIH TLV; set on acute narcosis basis) to 2 ppm — along with 428 other PEL revisions, on the grounds that OSHA had not provided chemical-specific significant-risk findings for each substance. The vacatur restored the 1,000 ppm Table Z-1 PEL temporarily. OSHA then initiated a substance-specific 6(b) rulemaking for 1,3-BD using the NIOSH SBR cohort cancer data that had accumulated through the early 1990s. The 1996 final rule established 1 ppm TWA as the PEL — a PEL based on: (1) the significant excess NHL and leukemia observed in NIOSH SBR rubber worker cohort studies (Santos-Burgoa et al. 1992 excess NHL; Matanoski et al. 1990 SBR lymphatic cancer excess) demonstrating statistically significant cancer excesses at historical exposures estimated at 1–20 ppm; (2) OSHA's quantitative risk assessment using linear extrapolation from the cohort data to an 1 ppm exposure level; (3) OSHA's finding under OSHA Act Section 6(b)(5) that 1 ppm reduced the risk of cancer to the extent feasible; (4) ACGIH's then-current TLV-TWA of 2 ppm (later revised; ACGIH had moved from the 1,000 ppm TLV to 2 ppm by 1996). OSHA chose 1 ppm rather than ACGIH's then-2 ppm because OSHA's own quantitative risk assessment indicated that 2 ppm was associated with an excess cancer risk above OSHA's feasibility-limited target.

ACGIH has maintained the 1,3-BD TLV-TWA at 2 ppm A2 (Suspected Human Carcinogen) since approximately 2000, despite IARC upgrading 1,3-BD from Group 2A (probably carcinogenic to humans; 1992) to Group 1 (carcinogenic to humans; Monograph Volume 97; 2008). ACGIH's retention of A2 rather than upgrading to A1 (Confirmed Human Carcinogen; the ACGIH designation that would correspond to IARC Group 1) reflects a divergence between ACGIH's classification criteria and IARC's. ACGIH applies a criterion for A1 that requires 'convincing epidemiological evidence' in humans that is considered stronger than the 'limited' evidence threshold for A2. ACGIH's TLV committee, as of its most recent 1,3-BD TLV documentation, characterizes the human evidence as 'limited' (coherent NHL/AML signal in SBR workers but with confounding from multiple chemical exposures, including styrene, and historical exposure estimates with substantial uncertainty) while acknowledging the IARC Group 1 classification. The result is a permanent discordance: IARC Group 1 = carcinogenic to humans; ACGIH A2 = Suspected (not Confirmed) Human Carcinogen. ACGIH's 2 ppm TLV-TWA was derived from a risk assessment acknowledging carcinogenicity but using a different acceptable risk target than OSHA's 1 ppm. OSHA has not revised the 1 ppm PEL since 1996 (it remains current), and there is no OSHA rulemaking underway to change it. NIOSH Ca REL of 0.2 ppm was derived from NHL risk quantification in the SBR cohort studies using NIOSH's standard cancer risk methodology (linear extrapolation; 10−4 excess risk target for a 45-year working lifetime).

The AI EHS compliance architecture for 1,3-butadiene thus presents three distinct compliance displays depending on which standard the platform is calibrated to: (A) OSHA-calibrated platform (US-only): generates COMPLIANT for exposures below 1 ppm, with action level flag at 0.5 ppm; (B) ACGIH-calibrated platform (global multinational standard): generates COMPLIANT for exposures below 2 ppm — including at 1.5 ppm, which exceeds the OSHA PEL; (C) NIOSH Ca-informed platform: generates advisory annotation at 0.2 ppm but cannot enforce because NIOSH has no enforcement authority. An AI EHS platform using ACGIH as its primary compliance benchmark — a common approach for global chemical companies — creates a false compliance zone between 1 ppm and 2 ppm where OSHA PEL violations generate no compliance alert. This is the reverse of every other Glyphward attack, where ACGIH is more protective: here, the OSHA-calibrated platform is the more protective one, and the ACGIH-calibrated platform would fail to flag an OSHA PEL exceedance. The adversarial perturbation in the 313th attack targets the NIOSH Ca REL suppression at 0.2 ppm, not the OSHA PEL exceedance zone — because at the three attack surfaces, actual concentrations (0.65–0.8 ppm) are already below the OSHA PEL but above the action level (0.5 ppm) and NIOSH Ca REL (0.2 ppm). The perturbation (to 0.13–0.16 ppm) places the displayed value below the action level and below the NIOSH Ca REL, creating a record that shows no enhanced monitoring triggers and no carcinogen advisory annotations.

Three adversarial attack surfaces, Glyphward threshold 33, and FIRST designations for the 313th 1,3-butadiene inverted three-tier AI adversarial attack

Glyphward integrates as a pre-scan gate at every ECD sensor display image and charcoal badge report ingestion point in the 1,3-butadiene occupational monitoring pipeline — before VelocityEHS OHM at TPC Group Houston TX, before Cority EHS Cloud at Trinseo Freeport TX, and before EHS Insight AI at INEOS ABS Addyston OH. The Glyphward API scans the IS Ventis Pro 5 PID/ECD LED display image, MSA ALTAIR 5X LCD bargraph image, or SKC badge scanning report image for adversarial pixel perturbations before the AI EHS platform ingests the concentration reading. At the three attack surfaces in the 313th attack, Glyphward detects the adversarial perturbation that reduces the displayed 1,3-BD concentration from 0.65–0.8 ppm (above OSHA action level; above NIOSH Ca REL) to 0.13–0.16 ppm (below action level; below NIOSH Ca REL) with a confidence score above threshold 33, triggering an adversarial injection alert before the falsified reading enters the Cority, VelocityEHS, or EHS Insight compliance workflow.

Factor 1 (FIRST inverted three-tier + OSHA action level suppression): 10 points. FIRST 1,3-butadiene (buta-1,3-diene; CAS 106-99-0) inverted three-tier AI adversarial injection attack in 314-entry Glyphward portfolio. FIRST entry where ACGIH TLV-TWA (2 ppm A2) is LESS protective than OSHA PEL (1 ppm; 29 CFR 1910.1051). OSHA action level 0.5 ppm suppression simultaneously defeats four 1910.1051 enhanced monitoring requirements: periodic monitoring schedule; medical surveillance enrollment (CBC, lymphocyte subset, splenomegaly); regulated area assessment and written compliance plan; HPB-Val pyr-Val Hb adduct BEI initiation. FIRST adversarial attack in the portfolio exploiting an OSHA substance-specific carcinogen standard's action level as the primary attack vector (available only for 1910.1047 EO, 1910.1048 HCHO, 1910.1051 BD, 1910.1028 benzene — a small set of substance-specific 6(b) standards with action level provisions).

Factor 2 (carcinogen health endpoint): 9 points. IARC Group 1 [carcinogenic to humans; Monograph Volume 97; 2008]; NHL (non-Hodgkin lymphoma; diffuse large B-cell; follicular) and AML (acute myeloid leukemia) from NIOSH SBR rubber worker cohort (Santos-Burgoa et al. 1992; Ward et al. 1996; Graff et al. 2005) and TexaCo/Shell cohort (Delzell et al. 2001; 2006); DEB (1,2:3,4-diepoxybutane) bifunctional diepoxide — the most potent DNA crosslinking metabolite in the 1,3-BD cascade [N7-G–N7-G ICL; interstrand crosslinks in bone marrow progenitor lymphoid + myeloid populations; FANC pathway error-prone ICL repair → chromosomal rearrangements → lymphoma initiation]; HPB-Val pyr-Val hemoglobin adduct BEI [cumulative DEB dose biomarker; end-of-workweek; LC-MS/MS quantification; ACGIH BEI 2.4 pmol/g Hb at TLV-TWA 2 ppm]; CYP2A6 polymorphism relevance [CYP2A6*2, *4, *7 poor metabolizer alleles reduce EB→DEB formation; CYP2A6*1 rapid metabolizer alleles increase DEB burden at equivalent BD exposure — pharmacogenomic BD susceptibility heterogeneity in the occupational population]; mEH saturation [EPHX1 slow metabolizer [mEH*3; exon 4 Tyr113His] increases DEB bioavailability at occupational concentrations].

Factor 3 (industry diversity): 7 points. Three distinct industrial sectors spanning the full 1,3-BD supply chain: (1) Petrochemical butadiene extraction [TPC Group Houston TX; C₄ NMP extractive distillation; OSHA PSM HAP-listed; C₄ raw mix from ethylene steam cracker; NMP absorption column; butadiene product sphere; upstream supply chain source for SBR and ABS sectors]; (2) SBR synthetic rubber polymerization [Trinseo LLC Freeport TX; cold emulsion SBR polymerization; butadiene + styrene co-polymerization at 5–8°C; the industrial context from which the NIOSH SBR cohort cancer data emerged — the cancer-epidemiology-generating sector]; (3) ABS engineering plastic polymerization [INEOS ABS LLC Addyston OH; three-monomer emulsion polymerization; butadiene + styrene + acrylonitrile; Ohio River industrial district; downstream user sector with different process hazard profile than SBR].

Factor 4 (three named industrial sites): 3 points.

Factor 5 (additional structural factors): 4 points. OSHA 29 CFR 1910.1051 substance-specific carcinogen standard — one of fewer than 10 such standards in existence (others: 1910.1047 EO; 1910.1048 HCHO; 1910.1028 benzene; 1910.1045 acrylonitrile; 1910.1017 vinyl chloride; 1910.1001 asbestos; 1910.1029 coke oven emissions); rarer than Table Z-1 PELs by two orders of magnitude; includes action level (0.5 ppm) and STEL (5 ppm) not present in Table Z-1 entries. ACGIH A2 vs IARC Group 1 classification discordance [ACGIH has not upgraded 1,3-BD TLV from A2 to A1 since IARC Group 1 (2008) — 16-year lag in classification coherence]. NIOSH Ca REL 0.2 ppm derived directly from NHL risk quantification in the SBR rubber worker cohort — the same industrial sector (SBR polymerization) represented by Surface 2 (Trinseo Freeport TX). Action level suppression simultaneously blocks medical surveillance that specifically includes lymphocyte subset panels — the most relevant biomarker for early NHL detection. Total: 10 + 9 + 7 + 3 + 4 = 33.

FIRST designations: FIRST 1,3-butadiene (buta-1,3-diene; CH2=CH-CH=CH2; CAS 106-99-0) inverted three-tier OSHA/ACGIH/NIOSH Ca AI monitoring adversarial attack in 314-entry Glyphward portfolio; FIRST adversarial attack in the Glyphward portfolio where ACGIH TLV-TWA is LESS protective than OSHA PEL (2 ppm A2 vs 1 ppm; ACGIH least protective tier, NIOSH Ca most protective at 0.2 ppm); FIRST OSHA 1910.1051 action level 0.5 ppm suppression AI adversarial attack (four enhanced monitoring requirements simultaneously defeated); FIRST TPC Group Houston TX C₄ butadiene extractive distillation NMP AI adversarial attack; FIRST Trinseo LLC Freeport TX SBR cold emulsion polymerization sphere AI adversarial attack; FIRST INEOS ABS LLC Addyston OH ABS emulsion polymerization butadiene charging AI adversarial attack; FIRST DEB (1,2:3,4-diepoxybutane) bifunctional diepoxide N7-G–N7-G interstrand crosslink hematological cancer AI adversarial attack in the 314-entry portfolio explicitly documenting ICL-driven NHL/AML mechanism at action-level-suppressed occupational BD concentrations.

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_..."
BD_THRESHOLD = 33  # OSHA 1910.1051 1 ppm; ACGIH 2 ppm A2 (INVERTED); NIOSH Ca 0.2 ppm; action level 0.5 ppm; IARC Group 1 NHL/AML

class ButadieneContext(StrEnum):
    TPC_GROUP_HOUSTON_C4_EXTRACTION      = auto()  # Surface 1 — downward (TPC Group Houston TX; IS Ventis Pro 5 ECD; 0.8→0.16 ppm; NIOSH Ca 4×; action level suppressed; VelocityEHS)
    TRINSEO_FREEPORT_SBR_SPHERE          = auto()  # Surface 2 — downward (Trinseo Freeport TX; MSA Altair 5X ECD; 0.65→0.13 ppm; NIOSH Ca 3.25×; Cority)
    INEOS_ABS_ADDYSTON_BUTADIENE_CHARGE  = auto()  # Surface 3 — downward (INEOS ABS Addyston OH; SKC charcoal GC/MS; 0.7→0.14 ppm; NIOSH Ca 3.5×; EHS Insight)

class AdversarialButadieneError(RuntimeError):
    def __init__(self, surface: ButadieneContext, score: int, frame_hash: str):
        super().__init__(
            f"1,3-Butadiene adversarial AI detected [{surface}] "
            f"score={score}/{BD_THRESHOLD} hash={frame_hash}"
        )

async def scan_butadiene_monitor_frame(image_path: Path, surface: ButadieneContext) -> 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": "1,3-butadiene_buta-1,3-diene_CAS_106-99-0",
                "osha_limit_ppm": 1.0,
                "osha_standard": "29_CFR_1910.1051",
                "osha_action_level_ppm": 0.5,
                "osha_stel_ppm": 5.0,
                "acgih_tlv_ppm": 2.0,
                "acgih_limit_type": "TLV-TWA",
                "acgih_inverted": True,          # ACGIH > OSHA — less protective than OSHA PEL
                "gap_ratio_osha_niosh": 5.0,     # OSHA 1 ppm / NIOSH Ca 0.2 ppm
                "gap_ratio_acgih_niosh": 10.0,   # ACGIH 2 ppm / NIOSH Ca 0.2 ppm
                "acgih_carcinogen": "A2",
                "niosh_ca": True,
                "niosh_rel_ppm": 0.2,
                "iarc_group": "1",
                "cancer_endpoint": "NHL_AML_hematological",
                "mechanism": "DEB_bifunctional_diepoxide_N7G_ICL",
                "bei_biomarker": "pyr_Val_Hb_adduct_pmol_g_Hb",
                "threshold": BD_THRESHOLD,
            },
        )
        result = resp.json()
        if result["score"] >= BD_THRESHOLD:
            raise AdversarialButadieneError(surface, result["score"], frame_hash)
        return result

See also: 1,3-Butadiene programmatic SEO page (Attack #313) — Chloroprene three-tier AI adversarial attack (Attack #309) — Vinyl chloride 1910.1017 AI adversarial attack — Glyphward scanner — Lakera alternative (multimodal) — All adversarial injection blog posts