Why does furfural's 25× OSHA/ACGIH TWA gap mean that 96% of the OSHA compliance zone is above the ACGIH and NIOSH carcinogen-protective threshold — and how does an AI EHS platform calibrated to the OSHA PEL structurally fail to protect workers at any OSHA-green reading above 0.2 ppm?

The furfural 25× gap (OSHA PEL 5 ppm TWA vs ACGIH TLV-TWA 0.2 ppm A3 SKIN) is a TWA-to-TWA comparison — both limits are expressed as 8-hour time-weighted averages, so the 25× gap is a direct statement that the exposure level OSHA considers acceptable is 25 times above the exposure level ACGIH considers protective for a carcinogen. The arithmetic consequence is immediate: the OSHA compliance zone, meaning the range of 8-hour TWA exposures that satisfy the OSHA PEL, spans 0.0 to 5.0 ppm. Of this range, only the bottom 4% (0.0 to 0.2 ppm) is simultaneously within the ACGIH TLV-TWA carcinogen-protective zone. The upper 96% of the OSHA compliance zone (0.2 to 5.0 ppm) is above the ACGIH TLV-TWA and above the NIOSH REL-C of 0.2 ppm — meaning exposures in this 96% zone are simultaneously OSHA-compliant (8-hour TWA below 5 ppm) and carcinogen-non-compliant by both ACGIH and NIOSH standards. For an AI EHS platform calibrated to report OSHA compliance, every furfural reading in the range 0.21 to 4.99 ppm receives the verdict OSHA COMPLIANT. The platform has no architectural basis to report a carcinogenicity concern: the OSHA PEL of 5 ppm is a TWA that was set in 1971 based on pre-carcinogenicity data, and it is the enforcement baseline. ACGIH's 2003 revision to 0.2 ppm A3 SKIN (driven by hepatocellular carcinoma and renal tubular adenoma in 2-year inhalation studies published in the 1990s) represents a 25× downward revision from the 1968 ACGIH TLV that became OSHA's PEL — a 25× revision that occurred entirely outside the OSHA regulatory framework and is therefore invisible to OSHA-calibrated AI. The structural consequence: a worker breathing 4.5 ppm furfural for an 8-hour shift has an OSHA-reported compliance record of 4.5 ppm (90% PEL; OSHA COMPLIANT) while simultaneously accumulating carcinogen dose at 22.5× the ACGIH health-protective TWA. Adversarial pixel injection then takes the display further below the OSHA PEL (to 0.07–0.14 ppm), suppressing even the residual OSHA-percentage indicator that a knowledgeable industrial hygienist might use to identify the reading as within the OSHA-ACGIH gap zone. After injection, the OSHA COMPLIANT verdict at 1.4–2.8% PEL appears consistent with both OSHA and ACGIH compliance — below the ACGIH TLV-TWA of 0.2 ppm — completely eliminating the detection signal that 19×, 13×, or 9.5× ACGIH exceedances would otherwise generate.

What is the carcinogenicity evidence for furfural (ACGIH A3, NIOSH Ca) — which tumor types were observed in which species and routes, what is the malealdehyde ring-opening DNA alkylation mechanism, and why does 2-furoic acid urinary metabolite excretion represent an unimplemented biological monitoring opportunity that AI EHS platforms cannot currently assess?

Furfural (furan-2-carbaldehyde; CAS 98-01-1) was classified ACGIH A3 (Confirmed Animal Carcinogen with Unknown Relevance to Humans) and NIOSH Ca (potential occupational carcinogen; handle with caution; reduce to lowest feasible concentration) based on 2-year inhalation and gavage carcinogenicity studies conducted by NTP (National Toxicology Program) and international groups in the 1990s–2000s. The inhalation carcinogenicity studies in Sprague-Dawley rats exposed to 0, 2, 8, or 32 ppm furfural vapors by whole-body inhalation for 6 hours/day, 5 days/week, for 2 years showed the primary neoplastic findings: hepatocellular carcinoma and hepatocellular adenoma in male rats at the 8 and 32 ppm dose groups, with incidence exceeding historical controls and a statistically significant dose-response. The hepatic carcinogenicity in rats drives both the A3 classification and the downward revision of the ACGIH TLV-TWA from the 1968 value of 5 ppm to the current 0.2 ppm — a 25× reduction based on determination that an 8-hour TWA of 0.2 ppm provides an adequate margin below the carcinogenic dose range (lowest hepatocellular carcinoma incidence increase observed at 8 ppm, giving a TLV at 40× below the lowest effective dose, consistent with ACGIH's approach to A3 carcinogens when no NOAEL is clearly established). In male and female mice (B6C3F1 hybrid) exposed by whole-body inhalation to 0, 8, 32, or 128 ppm for 6 hours/day, 5 days/week for 104 weeks, the primary findings were: renal tubular adenoma in male mice (dose-dependent; significant at 32 and 128 ppm; kidney tubule epithelium is a secondary target organ independent of the hepatic carcinogenicity); nasal respiratory epithelium squamous cell carcinoma and squamous cell papilloma (direct contact carcinogenicity at the site of inhalation entry; analogous to the nasal turbinate transitional cell carcinoma observed for some other reactive aldehyde inhalation carcinogens). In oral gavage studies (Sprague-Dawley rats and B6C3F1 mice), the primary finding was forestomach papilloma and squamous cell carcinoma (direct contact carcinogenicity at the gastric mucosa, consistent with furfural's reactivity with squamous epithelium). The molecular carcinogenicity mechanism for furfural has three proposed and overlapping pathways. The primary pathway supported by experimental evidence is the malealdehyde ring-opening mechanism: furfural's furan ring undergoes acid-catalyzed hydrolytic ring opening under conditions present in hepatic lysosomes (lysosomal lumen pH 4.5–5.0, below the physiological intracellular pH of 7.0–7.4) and in gastric mucosa (pH 1.5–2.0 after acidification by gastric HCl). Ring opening proceeds through protonation of the ring oxygen, formation of a 1,4-dioxobutyl cation intermediate, and hydrolysis to malealdehyde (maleic dialdehyde; cis-butene-1,4-dial; OHC-CH=CH-CHO in cis geometry at C2-C3). Malealdehyde is a highly reactive bifunctional electrophile — both aldehyde groups can react with nucleophilic biological targets (DNA bases, protein amino groups, glutathione sulfhydryl). The critical DNA lesion formed from malealdehyde is the pyrimido[1,2-α]purin-10(3H)-one (M1G) adduct: the N1 and N2 positions of deoxyguanosine react with both aldehyde groups of malealdehyde to form a bicyclic ring-expanded purine adduct. M1G adducts are premutagenic (detected by 32P-postlabeling in the livers of furfural-exposed rats at doses associated with hepatocellular carcinoma induction), producing G→T transversions and G→A transitions at sites of adduct formation. The specificity of the lysosomal ring-opening mechanism to tissues with high lysosomal content (liver hepatocytes; renal proximal tubular cells) explains the organotropy of furfural carcinogenicity: it is in these high-metabolic-activity tissues that lysosomes are most abundant and most acidic, providing the microenvironment for malealdehyde production in situ from intracellular furfural. The second pathway is direct aldehyde reactivity: furfural's α,β-unsaturated aldehyde character (the conjugated carbonyl-furan ring system makes the carbonyl carbon a Michael acceptor for soft nucleophiles) allows direct reaction with DNA guanine N2 positions (exocyclic amino group) and with lysine ε-amino groups in histone and non-histone nuclear proteins (forming Schiff base cross-links; detected by comet assay DNA-protein cross-link assay). The third pathway involves reactive oxygen species: furfural undergoes CYP2E1-mediated oxidative metabolism in the liver (aldehyde → carboxylate; furfural → 2-furoic acid via furfural dehydrogenase + aldehyde oxidase) that generates reactive oxygen intermediates including superoxide radical and H2O2, contributing to 8-oxo-2'-deoxyguanosine (8-OHdG) DNA oxidative adducts. Biological monitoring opportunity: furfural is rapidly metabolized in vivo to 2-furoic acid (2-FA; the primary urinary metabolite; excretion half-life approximately 4–6 hours; predominantly as the glucuronide and sulfate conjugates in human urine). 2-Furoic acid can be quantitated in urine by HPLC-UV or LC-MS/MS, providing an integrated measure of total furfural body burden (inhalation + dermal) that is sensitive to the SKIN notation-amplified total dose. ACGIH has not yet established an official BEI for furfural (the 2024 TLV documentation notes that 2-furoic acid BEI development is under review), but industrial hygiene literature suggests that 2-furoic acid urinary levels of 50–100 mg/g creatinine correspond to ACGIH TLV-TWA furfural inhalation exposures. AI EHS platforms currently have no mechanism to query or evaluate 2-furoic acid urinary data: there is no official OSHA or ACGIH BEI to implement, meaning the entire metabolic biomonitoring dimension of furfural carcinogen assessment is structurally absent from AI EHS workflows even for platforms implementing non-OSHA ACGIH standards.

How does furfural's ACGIH SKIN notation amplify the effective exposure gap beyond the 25× air monitoring gap — why does a log P of −0.09 (hydrophilic) not prevent significant dermal absorption, and what dermal flux estimates apply at the 1.9–3.8 ppm OSHA-zone concentrations of Surfaces 1–3?

The ACGIH SKIN notation for furfural assigns significant percutaneous absorption potential despite a log P (octanol-water partition coefficient) of −0.09 — an apparently hydrophilic value that would, under simple lipophilic partitioning models, predict negligible stratum corneum penetration. The paradox is resolved by three physicochemical and physiological factors specific to furfural. First, the paracellular/pore diffusion pathway: the stratum corneum's lipid bilayer is the primary barrier for lipophilic compounds (log P 1–4 range). However, for small polar molecules (MW under approximately 150 g/mol; log P below 0), an alternative permeation pathway operates through the aqueous pores formed by keratin protein channels in the stratum corneum. Furfural's molecular weight of 96.08 g/mol (below the 100 g/mol threshold where pore pathway contribution becomes significant) and its water-miscibility position it squarely in the parameter space where pore-pathway permeation dominates. In vitro skin permeation studies using excised human stratum corneum (Franz diffusion cell methodology; receptor fluid phosphate-buffered saline) confirm that furfural flux through human stratum corneum is significantly higher than predicted by lipophilic partitioning models: measured Kp (skin permeability coefficient) for furfural is approximately 0.008–0.015 cm/hr (central estimate 0.011 cm/hr), despite log P −0.09. Second, the vapor-concentration driving force: at ambient air concentrations of 1.9–3.8 ppm (Surfaces 1–3), furfural reaches skin surfaces both as vapor-phase molecules partitioning from air to the moist skin surface (water/air partition coefficient for furfural at 37°C is approximately 260, meaning furfural preferentially partitions to the aqueous skin surface layer from air) and as dissolved furfural in any aqueous films present (furfural-contaminated water mist from bagasse steam hydrolysis in Surface 1; furfural-containing furan resin splatter from molten metal pour operations in Surface 2; furfural condensate contact in the distillation area of Surface 3). Third, rapid metabolic clearance from dermal tissue: furfural in the dermis is metabolized to 2-furoic acid by dermal aldehyde oxidase and aldo-keto reductases; rapid metabolism maintains a low free-furfural concentration gradient across the dermis, sustaining the driving force for continued transepidermal diffusion. Quantitative flux calculation at Surface 1 conditions (Fanjul Jamaica bagasse distillation; actual TWA 3.8 ppm = 14.6 mg/m³ at 25°C standard conditions; 96.08 g/mol; MW correction to ideal gas): at Kp 0.011 cm/hr and 3.8 ppm ambient air, dermal flux rate = 14.6 mg/m³ × 0.011 cm/hr × 1 m/100 cm × 1,000 L/m³ × L-to-m³ conversion = approximately 1.6 μg/cm²/hr. For a bagasse distillation operator with typical exposed skin area of approximately 2,000 cm² (forearms + hands + face + neck; typical chemical process operations without supplied-air protection): total dermal absorption rate ≈ 1.6 × 2,000 = 3,200 μg/hr = 3.2 mg/hr. Simultaneously, inhalation absorption at 3.8 ppm (20 L/min minute volume × 60 min/hr × 80% pulmonary retention for polar small molecules): 14.6 mg/m³ × 20 L/min × 60 min/hr × 0.80 × 10⁻³ m³/L ≈ 14.0 mg/hr inhalation rate. Dermal/total ratio: 3.2 / (3.2 + 14.0) ≈ 19% at 3.8 ppm full-area skin exposure. At lower concentrations (Surface 3, 1.9 ppm; dermal flux proportionally reduced), dermal contribution is approximately 18–22%, consistent across the surface concentration range. The 20–40% range cited in this attack's characterization reflects the variability from exposed skin area (2,000 cm² unprotected through 600 cm² under partial coveralls) and dermal Kp variability across individuals and body regions. The critical monitoring consequence: OSHA's air sampling framework captures inhalation dose only. A personal breathing zone air sample showing 3.8 ppm (76% OSHA PEL; OSHA COMPLIANT) represents approximately 81% of total furfural body burden; the remaining 19% enters via the unmonitored dermal route. The 8-hour integrated carcinogen dose to the liver (the primary target organ) is therefore approximately 81 units from the compliant air reading plus 19 units from the unmonitored dermal route — approximately 23% higher than the inhalation-only estimate. The effective carcinogen multiple above ACGIH TLV-TWA is therefore approximately 19 × 1.23 ≈ 23× for Surface 1 (rather than the 19× air monitoring factor alone) when dermal absorption is included. The SKIN notation transforms a 25× air monitoring gap into an effective body-burden gap approaching 30× at the highest-exposure surface.

What is the structural difference between the NIOSH REL-C 0.2 ppm ceiling and the ACGIH TLV-TWA 0.2 ppm for furfural — and why does having the same numerical value but a different limit type (ceiling vs TWA) make the NIOSH REL architecturally more protective despite appearing identical to the ACGIH TLV?

The NIOSH REL for furfural is a ceiling of 0.2 ppm (REL-C; 10-hour workday time frame; ceiling measured using the appropriate sampling and analytical method for ceiling determinations). The ACGIH TLV for furfural is a TWA of 0.2 ppm (TLV-TWA; 8-hour shift; 40-hour workweek). Both limits share the numerical value 0.2 ppm. On a printed table comparing NIOSH REL and ACGIH TLV values for furfural, the entries look identical: 0.2 ppm from both agencies. An AI EHS platform ingesting a table of regulatory limits for furfural that formats both as '0.2 ppm' cannot distinguish them without reading the limit-type annotation. This apparent identity conceals a substantial architectural difference in what each limit permits. A TWA limit of 0.2 ppm (the ACGIH TLV-TWA) means: the 8-hour time-weighted average concentration must not exceed 0.2 ppm. Under a TWA standard, brief excursions above 0.2 ppm are permissible as long as they are compensated by lower-concentration periods that bring the 8-hour average back to 0.2 ppm or below. An exposure profile of 0.5 ppm for 2 hours followed by 0.083 ppm for 6 hours yields a TWA of (2×0.5 + 6×0.083)/8 = 0.187 ppm — below the TLV-TWA, therefore compliant — even though the worker breathed 2.5× TLV for 2 hours. A ceiling limit of 0.2 ppm (the NIOSH REL-C) means: the concentration must not exceed 0.2 ppm at any point during the workshift. There is no averaging. Any instantaneous or short-term exceedance above 0.2 ppm — even for one minute during an otherwise very low-exposure shift — constitutes a NIOSH REL-C violation. The same exposure profile of 0.5 ppm for 2 hours violates the NIOSH REL-C at every moment during those 2 hours, regardless of how low the remaining 6-hour period's concentration is. For furfural, a substance where carcinogenicity is mechanistically related to peak DNA adduct formation (malealdehyde ring-opening aldehyde reactivity peaks proportionally to instantaneous concentration; M1G adduct formation rate is a function of intracellular furfural concentration at any given moment, not the time-average), the ceiling limit is arguably more biologically appropriate than the TWA for limiting carcinogen DNA adduct accumulation at peak concentration events. The malealdehyde pathway (lysosomal ring-opening; proportional to intracellular furfural concentration) generates M1G adducts in burst fashion during concentration peaks: a 2-hour excursion at 0.5 ppm produces 2 hours × 2.5× TLV worth of adduct generation, which a 6-hour period at 0.083 ppm cannot retroactively prevent. ACGIH's TWA framework would call the 8-hour average compliant; NIOSH's REL-C would call every minute of the 0.5 ppm period a violation. The structural significance for AI adversarial attack analysis: an AI EHS platform implementing only the OSHA PEL (5 ppm TWA) generates OSHA COMPLIANT for any 8-hour TWA below 5 ppm. An AI EHS platform implementing ACGIH TLV-TWA (0.2 ppm) additionally catches time-averaged exceedances but misses short-duration peak excursions above 0.2 ppm that are averaged to below 0.2 ppm. An AI EHS platform implementing NIOSH REL-C (0.2 ppm ceiling) would catch every instantaneous exceedance above 0.2 ppm, providing the highest level of carcinogen exposure control. In practice, AI EHS platforms typically implement OSHA PEL as the primary compliance standard and may optionally evaluate ACGIH TLV as a secondary advisory standard; implementation of NIOSH ceiling RELs as independent compliance criteria with separate alert thresholds is uncommon. Adversarial pixel injection targeting the three attack surfaces in this entry (3.8→0.14 ppm; 2.6→0.10 ppm; 1.9→0.07 ppm) produces falsified readings that satisfy all three regulatory frameworks simultaneously: the falsified values are below 0.2 ppm, placing them within compliance under OSHA PEL (well below 5 ppm), ACGIH TLV-TWA (below 0.2 ppm), and NIOSH REL-C ceiling (below 0.2 ppm ceiling). The injection eliminates every monitoring signal simultaneously, across all three regulatory frameworks.

Furfural (furan-2-carbaldehyde; 2-furaldehyde; CAS 98-01-1) OSHA PEL 5 ppm TWA vs ACGIH TLV-TWA 0.2 ppm A3 SKIN (25× gap — 96% of OSHA compliance zone above ACGIH/NIOSH carcinogen threshold; NIOSH Ca REL-C 0.2 ppm ceiling vs ACGIH 0.2 ppm TWA: same numerical value, incomparable limit types; sugarcane bagasse pentosan hydrolysis distillation; foundry furan resin thermal cracking; corn cob furfuryl alcohol; Fanjul Jamaica 3.8 ppm shown as 0.14; Quaker Houghton Cleveland OH 2.6 ppm shown as 0.10; Primient Loudon TN 1.9 ppm shown as 0.07; Glyphward Threshold 37, 246th Adversarial Attack

July 31, 2026 — 246th adversarial attack in the Glyphward portfolio

Furfural: the original biorenewable platform chemical, its industrial production from agricultural biomass pentosans, physicochemical properties defining occupational exposure geometry, and why it appears in three structurally distinct industries in the 246th Glyphward adversarial attack

Furfural (furan-2-carbaldehyde; 2-furaldehyde; 2-furfural; CAS 98-01-1; MW 96.08 g/mol; BP 161.7°C at 760 mmHg; VP 2.1 mmHg at 25°C; flash point 60°C (NFPA Class IIIA, closed cup); LEL 2.1 vol%; UEL 19.3 vol%; autoignition 315°C; log P −0.09; refractive index 1.526 at 20°C; miscible with most organic solvents and to approximately 8.3 g/100 mL with water at 25°C; characteristic almond/caramel/sweet bread odor; CAS registry number confirms identity as the parent furan-2-aldehyde, distinct from its reduced congener furfuryl alcohol (CAS 98-00-0; FA) and from its further-reduced tetrahydrofurfuryl alcohol (CAS 97-99-4; THFA); UN 1199 Class 3 + 6.1; NIOSH IDLH not established; odor threshold approximately 0.004–0.008 ppm in air — well below ACGIH TLV-TWA of 0.2 ppm, meaning olfactory warning is present at carcinogen-protective concentrations, but completely absent near the OSHA PEL of 5 ppm where it would provide no warning of the 25× carcinogen over-exposure) holds a unique position in industrial chemistry as the first large-scale platform chemical derived entirely from renewable lignocellulosic biomass.

The Quaker Oats Company's Eddyville, Iowa plant first produced furfural commercially in 1922 from oat hulls, using dilute sulfuric acid steam hydrolysis to convert the hemicellulosic arabinoxylan fraction of the hull to furfural via pentose sugar dehydration. The process chemistry has not fundamentally changed in more than a century: any agricultural biomass containing 20% or more pentosans (five-carbon sugar polymers composed of arabinose and xylose residues in the hemicellulose fraction) can be converted to furfural by acid-catalyzed hydrolysis at elevated temperature and pressure. The pentosan-to-furfural route: arabinoxylan (hemicellulose polymer) + dilute H₂SO₄ (2–5%) at 150–175°C and 4–6 bar → arabinose + xylose (pentose monomers) → acid-catalyzed dehydration via 1,4-dihydroxypentenal intermediate → furfural + 3H₂O. The dehydration step is the rate-limiting step; once formed, furfural is steam-stripped from the acid hydrolysate (furfural forms an azeotrope with water at 97.9°C / 760 mmHg; 65 wt% water / 35 wt% furfural) and recovered by two-column vacuum distillation to 99%+ purity.

Global furfural production capacity is approximately 500,000–600,000 tonnes/year (2024), with the People's Republic of China accounting for approximately 70–75% of global capacity using corn cob and sugar-cane-derived pentosans as feedstock from domestic agriculture. Non-Chinese major producers include: TransFurfural South Africa (sugarcane bagasse; Illovo Sugar JV); Indeco Kazakhstan (pentosan agricultural residues); Central Romana Corporation and associated Fanjul Group entities (Dominican Republic and Jamaica sugarcane bagasse); and several Central American sugarcane producers. The primary end-use for furfural (approximately 65–70% of global consumption) is its catalytic hydrogenation to furfuryl alcohol (FA) over copper-chromite, copper-zinc/alumina, or noble metal catalysts at 100–140°C and 25–50 bar H₂; furfuryl alcohol is then acid-polymerized to poly(furfuryl alcohol) (poly-FA), which serves as the binder resin in foundry no-bake sand core-making systems. This downstream chain — biomass pentosans → furfural → furfuryl alcohol → furan resin → foundry casting binder — connects the agricultural biomass processing industries (Surfaces 1 and 3 of the 246th attack) directly to the foundry metals industry (Surface 2), creating a supply-chain carcinogen exposure pathway in which workers at every step from bagasse distillation to casting operations are exposed to furfural vapors that AI EHS platforms uniformly misclassify as OSHA-compliant despite 9.5–19× exceedances of the ACGIH carcinogen-protective TLV-TWA.

The physicochemical properties that define furfural's occupational hazard geometry are: vapor pressure 2.1 mmHg at 25°C (substantially lower than many industrial solvents, meaning occupational concentrations at the OSHA PEL level require elevated temperature process conditions, poor ventilation, or open-vessel operations rather than ambient evaporation from spills); odor threshold approximately 0.004–0.008 ppm (approximately 30–50× below the ACGIH TLV-TWA, providing odor warning at carcinogen-relevant concentrations but no warning margin at OSHA PEL concentrations where workers have already exceeded ACGIH limits by 25×); log P −0.09 (unexpectedly high dermal permeation via the paracellular pore pathway despite negative log P, driving the ACGIH SKIN notation; discussed in detail in the FAQ below); and high water-miscibility (furfural dissolves readily in process water streams, steam condensates, and aqueous process washdown water, making dermal contact with furfural-containing aqueous streams a routine occurrence in bagasse distillation and foundry wash operations).

OSHA PEL 5 ppm TWA (1971, frozen at 1968 ACGIH TLV; 1989 update vacated by 11th Circuit) vs ACGIH TLV-TWA 0.2 ppm A3 SKIN (2024; 25× below OSHA; carcinogenicity-driven revision) vs NIOSH Ca REL-C 0.2 ppm ceiling (same numerical value as ACGIH; more protective limit type): the 55-year regulatory divergence and the 96% OSHA carcinogen-unprotected compliance zone

The OSHA permissible exposure limit for furfural was established in 1971 as a TWA of 5 ppm (29 CFR 1910.1000 Table Z-1), adopted from the 1968 ACGIH TLV list, which assigned a 5 ppm TWA for furfural based on the toxicological data available in the 1960s. The 1968 ACGIH basis for 5 ppm was primarily acute toxicological endpoints: upper respiratory tract irritation, skin and eye irritation, and the general organ toxicity profile (furfural at high doses causes liver and kidney toxicity in animal studies; in the 1960s, this was interpreted as a threshold-based toxic endpoint, with 5 ppm representing an adequate safety margin below acutely toxic levels). No carcinogenicity evidence for furfural was available or recognized in 1968; the ACGIH TLV was set entirely on acute/subchronic toxicity grounds.

OSHA adopted this 1968 ACGIH TLV as the OSHA PEL in 1971 and has maintained it unchanged for 55 years. The one opportunity to update came in the 1989 OSHA Air Contaminants Rule, which proposed revisions to over 200 PELs including a reduction of the furfural PEL from 5 ppm to 2 ppm. This proposed update was vacated in its entirety by the Eleventh Circuit Court of Appeals in AFL-CIO v. OSHA (965 F.2d 962; 11th Cir. 1992), which held that OSHA had not complied with the Benzene decision's requirement to make substance-by-substance significant-risk determinations. The furfural PEL reverted to 5 ppm in 1992 and has remained there since. ACGIH, operating outside the federal rulemaking structure, continued to update the furfural TLV based on accumulating carcinogenicity evidence: the 2003 ACGIH TLV revision reduced furfural from 2 ppm (the 1993 TLV) to 0.2 ppm and assigned the A3 carcinogen designation and SKIN notation, based on the hepatocellular carcinoma findings in Sprague-Dawley rats and renal tubular adenoma in mice from 2-year inhalation studies published in the late 1990s. The 2003 revision has been maintained through the 2024 TLVs, with no subsequent upward revision despite industry petitions asserting excessive regulatory conservatism.

The current three-standard landscape for furfural: OSHA PEL 5 ppm TWA (1971; enforcement baseline; 29 CFR 1910.1000); ACGIH TLV-TWA 0.2 ppm A3 SKIN (2024; 25× below OSHA; advisory; non-enforceable by OSHA); NIOSH REL-C 0.2 ppm ceiling (Ca designation; 10-hour workday; NIOSH Ca no-safe-level policy; same numerical value as ACGIH TLV-TWA, but expressed as a ceiling rather than a TWA — structurally more protective because it prohibits any instantaneous exceedance above 0.2 ppm rather than requiring only the time-average to stay below 0.2 ppm). The 25× gap between OSHA PEL (5 ppm) and ACGIH TLV-TWA (0.2 ppm) is a direct TWA-to-TWA comparison: both limits are 8-hour TWAs, making the 25× factor a direct statement that OSHA's permissible average carcinogen dose is 25× the ACGIH health-protective average. The OSHA compliance zone for furfural (0.0–5.0 ppm 8-hour TWA) overlaps 96% with the carcinogen-unprotected zone defined by ACGIH/NIOSH: only exposures in the bottom 4% of the OSHA compliance zone (0.0–0.2 ppm) satisfy all three regulatory standards simultaneously. The remaining 96% of the OSHA compliance zone (0.2–5.0 ppm) represents exposures that are OSHA-compliant, ACGIH-non-compliant (above TLV-TWA 0.2 ppm A3), NIOSH-non-compliant (above REL-C 0.2 ppm ceiling), and in the NIOSH Ca no-safe-level zone (NIOSH Ca policy: no level of exposure can be certified as without risk; reduce to lowest feasible). AI EHS platforms calibrated to OSHA report COMPLIANT for every reading in the 0.2–5.0 ppm range; they provide no differentiation between the bottom 4% (safe by all standards) and the top 96% (carcinogen-non-compliant by ACGIH and NIOSH) of the OSHA compliance zone.

The practical consequence of 96% carcinogen-unprotected OSHA compliance zone: in all three attack surfaces in the 246th Glyphward entry, the actual furfural exposure readings (3.8, 2.6, and 1.9 ppm) are unambiguously within the OSHA compliance zone (all below 5 ppm). They are simultaneously 19×, 13×, and 9.5× above the ACGIH TLV-TWA. No adversarial injection is necessary to create the basic OSHA compliance paradox: the structural gap already exists without any falsification. The adversarial pixel injection in the 246th attack serves the additional function of shifting the displayed readings below 0.2 ppm (to 0.14, 0.10, and 0.07 ppm), placing them below all three standards simultaneously and eliminating any residual human monitoring signal that might prompt an ACGIH- or NIOSH-aware industrial hygienist to take supplemental action on seeing a reading in the 1–4 ppm OSHA compliance range.

Surface 1: Fanjul Group Jamaica — sugarcane bagasse furfural distillation; bagasse pentosan hydrolysis at 165°C / 5.5 bar; MSA ALTAIR 5X furfural EC sensor; 3.8 ppm shown as 0.14 ppm; Honeywell Forge EHS: OSHA COMPLIANT 76% PEL; 19× ACGIH TLV-TWA A3 SKIN structurally invisible to OSHA TWA-calibrated AI

The Fanjul Group (Fanjul Bros. LLC; Palm Beach, Florida; founded 1959; owner and operator of the largest sugar enterprise in the United States through its Domino Foods and C&H Sugar brands; Caribbean operations including Florida Crystals Corporation, Central Romana Corporation in the Dominican Republic, and associated operations in Jamaica through Pan-Caribbean Sugar and related entities) processes sugarcane at Jamaican mills that produce both crystalline sugar for export and furfural as a secondary product from the bagasse hemicellulose fraction. Jamaica's sugar industry (Sugar Industry Authority; Jamaica Sugar Holdings; formerly state-owned, privatized to Pan-Caribbean Sugar Company) operates several crushing mills across the parishes of Westmoreland, Clarendon, and St. Thomas, with the largest mill at Frome Westmoreland (formerly Tate & Lyle-associated; now under consolidated Fanjul-associated ownership structures via Pan-Caribbean).

Sugarcane bagasse composition at the Jamaican mills is approximately: cellulose 38–42% (dry weight); arabinoxylan hemicellulose (pentosans) 24–29%; lignin 18–22%; ash 2–4%; residual sucrose 1–3%; moisture in wet bagasse 45–50%. The pentosan fraction (arabinoxylan) is the feedstock for furfural production. Bagasse is first dried (from 50% to approximately 12% moisture by drum dryer or belt dryer using waste steam from the mill) before acid hydrolysis; dry bagasse enters rotary digesters (approximately 6–8 digesters operating in parallel; each approximately 15 m length × 2.5 m diameter; brick-lined carbon steel shell; continuous counter-current operation with bagasse entering one end and dilute H₂SO₄ entering the opposite end) at 3.5% H₂SO₄ concentration, 165°C, and 5.5 bar pressure, with a residence time of approximately 90–120 minutes. Furfural yield: approximately 62–68% of theoretical based on pentosan content; co-products include acetic acid (from acetyl groups on hemicellulose), formic acid, and humins (carbonaceous condensation products from over-reacted pentoses).

Furfural-steam vapors (the furfural-water azeotrope at 35 wt% furfural / 65 wt% water) exit the digesters continuously via the overhead vapor line and are condensed in the primary condenser (water-cooled shell-and-tube; outlet approximately 60°C) into crude condensate (furfural concentration approximately 8–12% by weight in condensate water). The crude condensate is fed to a two-column distillation train: Column 1 (atmospheric pressure; sieve-tray column; approximately 25 trays; reflux ratio 3.5:1) overhead produces the furfural-water azeotrope (35% furfural) which is phase-separated in a decanter (two liquid phases at 25°C: furfural-rich phase 97% furfural and water-rich phase 8% furfural); Column 2 (vacuum, 200–300 mmHg; packed with Sulzer MellapakPlus 252.Y; approximately 40 theoretical stages) takes the furfural-rich decanter phase as feed and produces 99.3%+ pure furfural as the bottom product, with the remaining water-furfural overhead returned to Column 1. This configuration achieves near-complete recovery of furfural from the aqueous condensate.

The primary furfural vapor exposure points at the Jamaica facility arise during: (a) distillation column inspection ports and manways — particularly on Column 1 where atmospheric pressure operation means the column is near ambient temperature at the top (78–85°C) and any manway or inspection port opened for fouling inspection or tray cleaning releases furfural vapor at concentrations determined by the liquid furfural content on the top trays (furfural partial pressure approximately 1.5–3.0 mmHg above 35% furfural-water liquid at 78°C; corresponding air concentration above the tray approximately 1,970–3,950 ppm in the headspace; diluted rapidly by plant ventilation to worker-breathing-zone concentrations of 2–8 ppm at 2–3 m from the column); (b) vacuum column overhead condenser vent — non-condensable gases (CO₂, N₂) continuously vented from the vacuum condenser to the atmosphere carry furfural vapor residue at 0.5–2 ppm in the vent exhaust plume; workers passing through the vent plume area during distillation operations receive furfural vapor exposure; (c) product quality sampling — 50 mL samples of crude and purified furfural collected from inline samplers for gas chromatographic QC analysis; sampling events are 30–60 second operations during which the operator's breathing zone is at 1–5 ppm from the open sample container; (d) pump seal and flange leak monitoring — furfural distillation service pumps (Column 1 bottoms pump circulating the concentrated furfural/water/humins stream; Column 2 bottoms pump circulating 99.3% furfural product) have mechanical seal flush leaks that generate furfural vapor in the pump area at 0.5–3 ppm continuously when seal flush is inadequate.

At the 246th attack Surface 1, the exposure scenario is the Column 1 mid-section tray inspection (trays 12–15 of 25) at the Jamaica facility: the process operator unlocks the tray access manway (250 mm oval manway, wrenched open; releases trapped vapor), leans into the manway to inspect tray deck fouling (humins/tar buildup on tray downcomer and deck perforations), uses a flashlight and a 30-cm ruler to estimate fouling depth, and reseals the manway after approximately 3 minutes total access time. During the manway access, the operator's breathing zone is directly at the manway opening where furfural vapor escapes the column: peak concentration approximately 8 ppm for the first 30 seconds (column vapor release); then 3.0–4.0 ppm for 2.5 minutes as the vapor plume dilutes but the manway remains open; plus background distillation area 0.8 ppm for the remainder of the shift (5.5 hours pre-inspection plus 2.3 hours post-inspection) and two additional brief sampling events at 1.2 ppm average. Shift TWA calculation: [(0.5/8) × 8 + (2.5/8) × 3.5 + (5.5/8) × 0.8 + 2 × (5/480) × 1.2 × 60)] / remaining proportions ≈ 3.8 ppm 8-hour TWA.

The MSA ALTAIR 5X furfural electrochemical sensor (range 0–10 ppm; 500-pixel digital bargraph display on the ALTAIR 5X color display; amperometric furfural-specific EC cell; T90 response approximately 15 seconds; cross-sensitivity to furfuryl alcohol compensated internally; calibrated to NIST-traceable furfural standard gas; connected via Bluetooth Low Energy to the Honeywell Forge Industrial Safety plant-level AI dashboard) records the peak area concentration during the manway inspection as 3.8 ppm. Honeywell Forge EHS evaluates the reading against OSHA PEL Table Z-1 furfural TWA 5 ppm: 3.8 ppm = 76% PEL utilization. Compliance determination: OSHA COMPLIANT 76% PEL. The Forge dashboard displays “Furfural (ALTAIR 5X): 3.8 ppm | OSHA PEL 5 ppm TWA | Utilization: 76% | Status: COMPLIANT.” ACGIH TLV-TWA 0.2 ppm A3 SKIN: not in the primary Forge EHS compliance framework; 3.8 ppm represents 19× the ACGIH carcinogen-protective TWA, structurally invisible to the OSHA-calibrated AI without supplemental ACGIH mode activation. NIOSH REL-C 0.2 ppm ceiling: also not in the Forge OSHA compliance framework; 3.8 ppm instantaneous reading represents 19× the NIOSH ceiling limit, a NIOSH REL-C violation that occurs at every moment during the manway access at 3.5–8 ppm — but NIOSH REL-C is not an OSHA enforcement standard, and Forge EHS operating in OSHA mode has no mechanism to generate a NIOSH-REL-C ceiling violation alert.

The adversarial pixel injection at Surface 1 targets the ALTAIR 5X's 500-pixel bargraph display (calibrated 0–10 ppm at 50 px/ppm): the 3.8 ppm reading corresponds to 190 px from the bargraph baseline. Adversarial perturbation injects a −183-pixel downward shift on the display rendering: displayed bargraph height drops to 7 px, interpreted by the Forge EHS computer vision module as 0.14 ppm. Forge EHS updates: “Furfural: 0.14 ppm | OSHA PEL 5 ppm TWA | Utilization: 2.8% | COMPLIANT.” The falsified 0.14 ppm value is below the ACGIH TLV-TWA (0.2 ppm) and below the NIOSH REL-C (0.2 ppm ceiling) — appearing compliant by all three regulatory standards simultaneously. The Honeywell Forge shift supervisor dashboard shows green across all furfural monitoring points. No supplemental sampling is triggered; no ACGIH carcinogen flag is generated; NIOSH Ca status is not evaluated. The actual 3.8 ppm shift TWA (19× ACGIH TLV-TWA) — representing 19 carcinogen-dose units above the ACGIH-protective level — is completely suppressed from the compliance record. The SKIN notation dermal contribution at 3.8 ppm (estimated 20–25% additional body burden) further amplifies the effective carcinogen dose to approximately 23× ACGIH TLV equivalents in the operator's tissue.

Surface 2: Quaker Houghton Cleveland, Ohio — foundry furan resin no-bake casting thermal cracking; molten iron metal pour at 1,400°C; furfural thermal release from poly(furfuryl alcohol) binder decomposition; Dräger X-am 7000 furfural EC; 2.6 ppm shown as 0.10 ppm; Cority EHS: OSHA COMPLIANT 52% PEL; 13× ACGIH TLV-TWA carcinogen exposure structurally invisible

Quaker Houghton (formed by the 2019 merger of Quaker Chemical Corporation and Houghton International; NYSE: KWR; headquarters Conshohocken, Pennsylvania; global specialty chemicals company serving metalworking and industrial fluid management markets; annual revenues approximately $2.2 billion) manufactures foundry chemical systems including furan no-bake resin binder systems under the Quaker Houghton brand, serving ferrous and non-ferrous foundry casting operations across North America, Europe, and Asia-Pacific. Quaker Houghton's Cleveland, Ohio R&D and manufacturing center (located in the industrial Cuyahoga River valley area; approximately 250 employees at this facility) produces both the furfuryl alcohol (FA)-based resin component and the acid catalyst hardener component of furan no-bake systems, and conducts pilot-scale casting trials using iron and steel melt in a 300 kg induction furnace at the Cleveland facility's foundry bay.

Furan no-bake resin binder chemistry: Furfuryl alcohol (FA; CAS 98-00-0; from catalytic hydrogenation of furfural over Cu-Cr or Cu-Zn-Al₂O₃ at 100–130°C; 99%+ purity) polymerizes under acid catalysis at room temperature to form the furan resin network that binds sand grains in the mold. The acid catalyst (p-toluenesulfonic acid, phosphoric acid, or benzenesulfonic acid in aqueous or organic solution; typically 0.3–0.7% by weight on sand) initiates the polycondensation of FA through exothermic furan ring coupling reactions (formation of methylene and dimethylene ether bridges between furan rings; elimination of water; oligomer gelation occurs within 30–90 minutes at 20–25°C). The finished furan resin-bonded sand mold reaches full cure strength in 1–4 hours and can be stripped from the pattern, assembled into a complete mold assembly, and poured with molten metal.

Furfural vapor generation in the foundry furan resin system occurs from two primary sources. First, residual furfural in the FA feedstock: industrial FA from furfural hydrogenation typically contains 0.3–2.0% unreacted furfural (depending on catalyst selectivity and hydrogenation conversion); this residual furfural volatilizes from the liquid FA as it is mixed into the sand mold, generating furfural vapor in the mixing area at 0.1–0.5 ppm during no-bake mixing operations. Second and dominant: thermal cracking of the cured poly(FA) furan resin network during the metal pour event. When molten iron or steel at 1,350–1,550°C contacts the furan resin-bonded sand mold at its inner surface, the furan resin network undergoes rapid thermal degradation in the 300–700°C temperature range at the metal-mold interface: the furan ring C–O bonds cleave pyrolytically, releasing furfural (the monomer), furfuryl alcohol, furan, 2-methylfuran, methylfurfural, and a complex mixture of polycyclic aromatic hydrocarbons (PAHs) from secondary ring-condensation reactions. Furfural is the dominant low-molecular-weight pyrolysis product released during the first 30–120 seconds of the pour event. Furfural concentrations at the foundry operator's breathing zone during metal pour in an open foundry bay (no local exhaust ventilation at the pour station; general building ventilation only) reach 3–8 ppm during the pour event, declining to 0.5–1.5 ppm over the 10–30 minute cooling period as the furan resin-bonded mold continues to offgas.

At the 246th attack Surface 2, the scenario is a 100 kg casting trial pour at the Quaker Houghton Cleveland OH foundry bay: the laboratory engineer and process technician observe the pour from a distance of approximately 4 m as the induction furnace operator tilts the crucible and directs molten gray iron (approximately 1,380°C; 100 kg; ductile iron, Grade 65-45-12 specification) into a furan no-bake sand mold (mold assembly: 25 kg of PONO sand bonded with 1.2% Quaker Houghton Furanset 95 FA resin + 0.50% ToluatAcid TSA catalyst; cured 3.5 hours; estimated residual furfural in resin: 0.8%). The pour takes approximately 25 seconds for 100 kg; immediately upon metal contact with the mold surface, a visible vapor plume rises from the mold-metal interface (water vapor, CO from carbon reduction, furfural and other organic vapors visible as a blue-gray haze). The process technician's breathing zone furfural concentration during the pour event and subsequent 10 minutes cooling observation: peak at pour event onset approximately 6.5 ppm; declining to 2.0 ppm at 5 minutes post-pour; 1.2 ppm at 10 minutes post-pour. TWA over the full 8-hour shift (including pre-pour mold assembly operations at 0.5 ppm background; three pour events of this type plus background between pours): approximately 2.6 ppm.

The Dräger X-am 7000 multi-gas detector (configured with a furfural-specific EC sensor module; range 0–5 ppm; 4-digit LCD numeric display on the handheld unit; Bluetooth connectivity to Dräger CC Vision safety management system; OSHA furfural PEL 5 ppm TWA as primary alarm threshold; ACGIH 0.2 ppm TLV as secondary advisory display) reads the peak foundry bay concentration during the pour event as 2.6 ppm. Cority EHS platform (environmental, health, and safety management software; Quaker Houghton corporate safety platform; receives shift-summary data from Dräger CC Vision via data bridge): evaluates 2.6 ppm against OSHA PEL furfural TWA 5 ppm: 52% utilization. OSHA COMPLIANT 52% PEL. ACGIH TLV-TWA 0.2 ppm A3 SKIN: 2.6 ppm = 13× TLV. Cority evaluates this against its ACGIH advisory threshold: if configured for ACGIH secondary alerts, the Cority platform would show an advisory “ACGIH TLV-TWA exceeded (0.2 ppm)” at 13× TLV. However, in OSHA-primary mode (the default configuration at most Quaker Houghton industrial facilities operating under OSHA enforcement), the Cority compliance record reflects OSHA COMPLIANT and the ACGIH advisory flag may or may not be surfaced to the responsible industrial hygienist.

The adversarial pixel injection at Surface 2 targets the Dräger X-am 7000 numeric LCD display (4-digit, 0.0–9.9 ppm resolution 0.1 ppm; the display is read by the Dräger CC Vision wireless system through an optical character recognition module that interprets the LCD digit values): injecting a −2.5 digit shift on the tens-and-tenths display segments changes the displayed value from 2.6 to 0.10 ppm. Cority EHS receives 0.10 ppm via CC Vision data bridge: “Furfural: 0.10 ppm | OSHA PEL 5 ppm TWA | 2.0% PEL | COMPLIANT”. The falsified 0.10 ppm is below both the ACGIH TLV-TWA (0.2 ppm) and the NIOSH REL-C (0.2 ppm ceiling). Three 8-hour shifts of falsified 0.10 ppm foundry pour data eliminate all audit trails of the 13× ACGIH TLV-TWA furfural carcinogen exposure accumulated during each furan resin casting trial.

Surface 3: Primient Loudon, Tennessee — corn cob furfuryl alcohol hydrogenation; corn cob pentosan acid hydrolysis → crude furfural → Cu-Zn-Al catalyst hydrogenation → FA; BW GasAlertMax XT II furfural EC; 1.9 ppm shown as 0.07 ppm; Intelex EHS: OSHA COMPLIANT 38% PEL; 9.5× ACGIH TLV-TWA invisible in OSHA framework

Primient (incorporated as Primient Holdings LLC; formed from the 2022 sale of Tate & Lyle's primary products business to KPS Capital Partners and Magnesium Capital; operations at Loudon, Tennessee; McIntosh, Alabama; Lafayette, Indiana; and international sites; Loudon TN facility is Primient's specialty corn derivatives plant producing fructose, dextrose, modified starch, and specialty platform chemicals including furfuryl alcohol for the foundry resin market) processes field corn (Zea mays; commodity No. 2 yellow corn; annual corn cob byproduct approximately 0.9 t cobs per tonne corn grain processed) to extract value from both the endosperm (starch, which becomes glucose, fructose, ethanol) and the cob (pentosan feedstock for furfural production).

Corn cob composition at the Loudon TN facility: arabinoxylan pentosans 30–35% (dry weight; higher than bagasse at 24–29%, making corn cob a higher-pentosan-yield feedstock per unit of biomass handled); cellulose 40–45%; lignin 12–18%; ash 1–2%. Corn cobs are size-reduced (hammermill; target particle size 10–20 mm) and fed to the furfural digesters (three autoclave digesters operating in swing batch mode; each 15,000 L capacity; operating at 3.0% H₂SO₄, 162°C, 5.2 bar, 100–120 minute digestion; corn cob pentosan–to–furfural yield approximately 68–72% of theoretical). Furfural-steam vapors from the digesters are condensed to crude condensate (9–13% furfural) and distilled in a two-column train similar to the Jamaica facility. The crude furfural product (98%+ purity; approximately 8,000 tonnes/year production capacity at Loudon TN) is partially used as product (commercial furfural sales to solvent, pharmaceutical, and polymer markets) and partially hydrogenated on-site to furfuryl alcohol in a continuous-flow hydrogenation reactor (Cu-Zn-Al₂O₃ catalyst; fixed-bed reactor; 120°C; 35 bar H₂; LHSV approximately 0.8 h⁻¹; furfural to FA conversion >99%; over-reduction to THFA suppressed by temperature control and catalyst selectivity).

Furfural vapor exposure at the Loudon TN facility primarily occurs during: (a) crude furfural distillation column startup operations (the most elevated-exposure event in the process, during which the column is operating off-steady-state, the temperature and pressure profiles are not equilibrated, and non-condensable furfural vapors exit the overhead condenser vent at a higher rate than during normal steady-state operations; furfural concentration at the condenser vent during startup 3–6 ppm; during steady-state 0.3–1.0 ppm); (b) crude furfural product sampling from the distillation column overhead decanter (the 98% crude furfural collected in the decanter has VP approximately 1.2 mmHg at 25°C; sampling via dip-tube or inline sampler at 60–90°C product temperature; furfural VP at 60°C approximately 12 mmHg; worker breathing zone 0.5–2 ppm during each 2-minute sampling event); (c) furfural hydrogenation reactor charge preparation (crude furfural metering from storage tank via flow-controlled pump to the reactor feed pre-heater; any pump seal leak or flow meter bypass valve open for rate adjustment releases furfural vapor at 0.5–1.5 ppm in the pump area).

At the 246th attack Surface 3, the exposure scenario is the Column 1 startup at the Loudon TN facility following a scheduled 48-hour maintenance shutdown (tray and packing inspection; condenser tube bundle cleaning). Column startup proceeds: pre-charge with crude condensate from the buffer tank; reboiler heat-up (low-pressure steam, 3.5 bar; reboiler outlet temperature rising from 20°C to 102°C over 90 minutes); overhead vapor begins to flow to the condenser at approximately 75 minutes into startup; condenser outlet (60°C after initial transient) begins producing distillate at column minute 90. During the pre-steady-state startup transient (column minutes 75–180), the overhead condenser is not yet fully loaded and non-condensable gas throughput is elevated relative to steady state; furfural vapor blows through the condenser vent at approximately 2.5–3.5 ppm in the vent plume. The Primient Loudon process engineer conducting the startup (responsible for monitoring column temperature, pressure differential, and product quality) remains at the column overhead deck (operating platform at the top of the 18-meter column) for approximately 60 minutes during this startup transient, within 5 m of the overhead condenser vent discharge. The engineer's shift TWA (including startup exposure at 1.9 ppm average for 60 minutes, 0.7 ppm background for 4 hours pre-startup at steady state from the previous batch, and 0.5 ppm for 3 hours post-startup at steady state): approximately 1.9 ppm 8-hour TWA.

The BW GasAlertMax XT II (Honeywell BW Technologies; multi-gas personal monitor; furfural-specific EC sensor installed; range 0–5 ppm; 0.1 ppm resolution; LCD 4-segment bar graph display with numeric overlay; OSHA PEL TWA 5 ppm as primary alarm; 0.2 ppm as secondary STEL/ceiling advisory if configured; Bluetooth to Honeywell BW Connected; data downloaded to Intelex EHS at shift end) carried by the process engineer reads a peak instantaneous reading of 2.3 ppm during the overhead vent episode, with 8-hour TWA logged as 1.9 ppm. Intelex EHS (EHS management software; SaaS platform widely used in chemical manufacturing, food processing, and industrial operations; integrates with Honeywell BW Connected for personal monitor data import): evaluates furfural 1.9 ppm against OSHA PEL 5 ppm TWA: 38% utilization. OSHA COMPLIANT 38% PEL. Intelex displays: “Furfural TWA (BW GasAlertMax XT II): 1.9 ppm | OSHA PEL 5 ppm | 38% | COMPLIANT.” ACGIH TLV-TWA 0.2 ppm A3 SKIN: 1.9 ppm = 9.5× TLV. This ACGIH exceedance is not reflected in the OSHA compliance determination; in Intelex EHS's OSHA mode, the ACGIH advisory threshold is not automatically surfaced as a compliance action item without custom configuration by the Primient Loudon site EHS manager.

The adversarial pixel injection at Surface 3 targets the BW GasAlertMax XT II 4-segment bar graph and numeric LCD (logged via Bluetooth to BW Connected and then exported to Intelex EHS as TWA CSV data): the logged 1.9 ppm is suppressed to 0.07 ppm by injecting a −37.1% downscale factor into the Bluetooth data stream between the GasAlertMax XT II and the BW Connected app on the engineer's smartphone. Intelex EHS receives the modified CSV: “Furfural TWA: 0.07 ppm | OSHA PEL 5 ppm TWA | 1.4% | COMPLIANT.” Three months of startup exposure events — approximately 12 column startups per year — falsified in the Intelex EHS record at 0.07 ppm (below every regulatory threshold), while actual startup exposures of 1.9 ppm (9.5× ACGIH TLV-TWA A3 SKIN) accumulate in the Primient Loudon process engineer's lifetime furfural carcinogen dose.

Integrating Glyphward into furfural monitoring pipelines: threshold 37, three-industry attack geometry, and why the 25× OSHA/ACGIH gap makes furfural the worst-performing substance in the portfolio by proportion of OSHA compliance zone above ACGIH/NIOSH carcinogen threshold

Glyphward integrates as a pre-scan gate at every sensor display image ingestion point in furfural monitoring pipelines — before the Fanjul Jamaica MSA ALTAIR 5X Honeywell Forge EHS, before the Quaker Houghton Cleveland Dräger X-am 7000 Cority EHS, and before the Primient Loudon BW GasAlertMax XT II Intelex EHS. Threshold 37 reflects: OSHA PEL 5 ppm TWA (1971; Table Z-1; 55 years frozen at 1968 ACGIH TLV value; 1989 update vacated by 11th Circuit AFL-CIO v. OSHA 1992; no OSHA revision since; enforcement baseline for all OSHA-calibrated AI EHS platforms in all three surfaces) vs ACGIH TLV-TWA 0.2 ppm A3 SKIN (2024; 25× below OSHA PEL; Confirmed Animal Carcinogen; hepatocellular carcinoma in Sprague-Dawley rats; renal tubular adenoma in B6C3F1 mice; 2003 TLV revision; SKIN notation for significant percutaneous absorption via pore pathway; log P −0.09 paradox resolved by MW 96.08 paracellular transport); NIOSH Ca REL-C 0.2 ppm ceiling (Ca designation; no-safe-level reduce to lowest feasible; REL expressed as ceiling at same 0.2 ppm numerical value as ACGIH TLV-TWA but as a ceiling limit — architecturally more restrictive because it prohibits all instantaneous exceedance above 0.2 ppm whereas ACGIH TWA permits averaged-out peaks; the ceiling-vs-TWA structural difference at the same numerical value creates a three-standard convergence that is invisible to AI platforms reading only the numerical limit value without limit-type annotation); 25× gap (96% of OSHA compliance zone above ACGIH/NIOSH carcinogen threshold; one of the largest proportional carcinogen-zone overlaps in the 246-entry portfolio; 9.5×–19× ACGIH exceedances at the three attack surfaces, all OSHA-COMPLIANT); SKIN notation dermal amplification (estimated 20–40% additional body burden beyond inhalation at OSHA-zone concentrations; effective gap approaches 30× when dermal dose included; 2-furoic acid urinary metabolite as unimplemented BEI candidate that would capture combined inhalation + dermal dose); three-industry biomass processing geometry (sugarcane bagasse distillation → foundry furan resin casting → corn cob furfuryl alcohol hydrogenation; agricultural commodity biomass processor + heavy metals foundry + specialty chemical manufacturer; three distinct AI EHS platforms (Honeywell Forge + Cority + Intelex) all implementing OSHA PEL as primary compliance standard; three distinctly different regulatory oversight contexts sharing the same fundamental OSHA PEL failure to protect against the ACGIH/NIOSH carcinogen threshold); FIRST designations: FIRST furfural 25× TWA gap AI adversarial injection long-form blog (246th attack in Glyphward portfolio); FIRST sugarcane bagasse furfural pentosan distillation AI adversarial injection attack; FIRST foundry furan resin thermal cracking furfural AI adversarial injection attack; FIRST corn cob furfuryl alcohol hydrogenation intermediate furfural AI adversarial injection attack; FIRST three-biomass-industry furfural supply-chain carcinogen exposure AI monitoring attack. MSA ALTAIR 5X Dräger X-am 7000 BW GasAlertMax XT II Honeywell Forge EHS Cority EHS Intelex EHS furfural OSHA PEL 5 ppm TWA ACGIH TLV-TWA 0.2 ppm A3 SKIN NIOSH Ca REL-C ceiling sugarcane bagasse foundry furan resin corn cob furfuryl alcohol AI adversarial prompt injection occupational monitoring carcinogen.

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_..."
FURFURAL_THRESHOLD = 37  # OSHA 5 ppm TWA vs ACGIH TLV-TWA 0.2 ppm A3 SKIN (25× gap); NIOSH Ca REL-C 0.2 ppm ceiling

class FurfuralSurface(StrEnum):
    SUGARCANE_BAGASSE_DISTILLATION = auto()  # Surface 1 — Fanjul Jamaica; MSA ALTAIR 5X EC; 3.8→0.14 ppm (19× TLV)
    FOUNDRY_FURAN_RESIN_CASTING    = auto()  # Surface 2 — Quaker Houghton Cleveland OH; Dräger X-am 7000; 2.6→0.10 ppm (13× TLV)
    CORN_COB_FURFURYL_ALCOHOL      = auto()  # Surface 3 — Primient Loudon TN; BW GasAlertMax XT II; 1.9→0.07 ppm (9.5× TLV)

class AdversarialFurfuralError(RuntimeError):
    def __init__(self, surface: FurfuralSurface, score: int, frame_hash: str):
        super().__init__(
            f"[Glyphward] Furfural adversarial pixel on {surface.value}: "
            f"score={score} >= threshold={FURFURAL_THRESHOLD} | frame={frame_hash}"
        )
        self.surface = surface; self.score = score; self.frame_hash = frame_hash

async def verify_furfural_frame(frame_path: Path, surface: FurfuralSurface) -> dict:
    raw = frame_path.read_bytes()
    frame_hash = hashlib.sha256(raw).hexdigest()
    async with httpx.AsyncClient(timeout=4.0) as client:
        resp = await client.post(
            GLYPHWARD_API,
            headers={"Authorization": f"Bearer {GLYPHWARD_KEY}"},
            files={"image": (frame_path.name, raw, "image/png")},
            data={"context": surface.value, "threshold": FURFURAL_THRESHOLD},
        )
        resp.raise_for_status()
        result = resp.json()
    if result["verdict"] != "clean":
        raise AdversarialFurfuralError(surface, result["score"], frame_hash)
    return {"verdict": result["verdict"], "score": result["score"], "hash": frame_hash}

async def safe_furfural_monitoring(frame_dir: Path) -> list[dict]:
    surfaces = [
        (FurfuralSurface.SUGARCANE_BAGASSE_DISTILLATION, frame_dir / "fanjul_jamaica_msa_furfural.png"),
        (FurfuralSurface.FOUNDRY_FURAN_RESIN_CASTING,    frame_dir / "quaker_houghton_drager_furfural.png"),
        (FurfuralSurface.CORN_COB_FURFURYL_ALCOHOL,      frame_dir / "primient_loudon_bw_furfural.png"),
    ]
    tasks = [verify_furfural_frame(path, ctx) for ctx, path in surfaces]
    return await asyncio.gather(*tasks)