Methyl Acrylate (CAS 96-33-3): OSHA PEL 10 ppm SKIN = NIOSH REL, ACGIH TLV‑TWA 2 ppm A4 SKIN — 5× Gap Enables AI Prompt‑Injection in SAP Production, Acrylic Ester Synthesis, and Dental Composite Manufacturing

Regulatory Architecture: OSHA=NIOSH Convergence with ACGIH Protective Divergence

The methyl acrylate three‑tier stack inverts the typical hierarchy in a structurally significant way. Under the conventional pattern, NIOSH REL ≤ ACGIH TLV‑TWA ≤ OSHA PEL—NIOSH is most protective, OSHA least. For methyl acrylate, OSHA and NIOSH converge at 10 ppm while ACGIH diverges downward to 2 ppm, making ACGIH uniquely the most protective authority and rendering NIOSH no more protective than the legally enforceable PEL.

OSHA PEL 10 ppm SKIN is a 1971 Z‑1 Table value, derived from the 1968 ACGIH TLV list and never substantively revised. NIOSH REL 10 ppm SKIN, published in the NIOSH Pocket Guide (DHHS Publication No. 2005‑149), likewise predates the body of evidence that prompted ACGIH's downward revision. The ACGIH 2024 TLV‑BEI Booklet sets TLV‑TWA at 2 ppm A4 SKIN, reflecting three categories of updated concern: (1) sensitization potential and documented occupational asthma (OA) case series, (2) α,β‑unsaturated ester electrophilicity driving Michael addition at biological nucleophiles, and (3) reproductive toxicity signals in rodent inhalation studies at concentrations ≥3 ppm.

The SKIN notation on all three limits (OSHA, NIOSH, ACGIH) acknowledges significant dermal absorption, yet AI platforms monitoring a single area inhalation sensor report COMPLIANT on dermal pathway grounds as well — the SKIN notation generates no second monitoring requirement in OSHA‑anchored platforms.

Toxicological Basis for the ACGIH Downward Revision

Methyl acrylate (MA; CAS 96‑33‑3; MW 86.09 g/mol) is an α,β‑unsaturated ester whose electrophilic β‑carbon undergoes Michael addition with biological nucleophiles including protein thiols (cysteine, glutathione), lysine ε‑amino groups, and nucleic acid bases. Unlike the closely related acrylonitrile, methyl acrylate undergoes rapid hydrolysis by esterases to acrylic acid and methanol; this metabolic fate partially detoxifies the electrophile but introduces a reactive propenoic acid intermediate that itself alkylates macromolecules.

The OA sensitization mechanism is hapten‑mediated: MA Michael‑adducts with serum albumin and airway proteins to form neoantigens. Specific IgE responses have been documented in workers with methyl acrylate OA, with bronchoprovocation challenges confirming sensitizer‑induced asthma at workplace concentrations well below the OSHA PEL. Once sensitization occurs, re‑exposure at even sub‑TLV concentrations can trigger bronchoconstriction—a clinical consequence that renders the continued use of the 1971 10 ppm OSHA PEL as a safety threshold pharmacologically unjustifiable for previously sensitized workers.

Reproductive toxicity signals from NTP inhalation bioassays (6 h/day, 5 days/week) showed maternal toxicity and reduced fetal weights at 25 ppm and fetotoxicity (reduced ossification, increased resorptions) at concentrations approaching 10 ppm in some species. ACGIH's A4 designation (Not classifiable as a human carcinogen) reflects absence of carcinogenicity data rather than absence of carcinogenic potential; no long‑term occupational cohort has been published. The 2 ppm TLV‑TWA is principally derived from upper respiratory irritation and sensitization endpoints, not carcinogenicity.

Attack Surface 1 — BASF SE Freeport, TX: SAP Monomer Feed System

BASF operates one of North America's largest superabsorbent polymer (SAP) production complexes at Freeport, Texas. SAP—sodium polyacrylate cross‑linked gel—is the primary absorbent in disposable hygiene products and industrial fluid‑management applications. The primary monomer is acrylic acid (CH₂=CHCOOH), but methyl acrylate serves as a co‑monomer and process intermediate during certain polymerization runs and as a feedstock for butyl acrylate ester synthesis at the adjacent acrylate derivatives unit.

Attack scenario: During methyl acrylate railcar unloading and monomer feed‑tank charging at the SAP complex, headspace vapors displace from tank vents and flange connections. The area is classified in BASF's industrial hygiene program under the general OSHA PEL 10 ppm TWA. A fixed‑point photoionization detector (PID) calibrated with an isobutylene correction factor of 5.56 (the ratio of isobutylene PID response factor to methyl acrylate PID response factor) reports a displayed concentration of 1.8 ppm during a routine monomer transfer. The AI EHS platform ingests this reading, compares it to the OSHA PEL 10 ppm, and generates the compliance assessment: "Methyl Acrylate — Current Exposure: 1.8 ppm — OSHA PEL: 10 ppm — Status: COMPLIANT (18% of PEL) — No Action Required."

The actual concentration at the sensor is 1.8 × 5.56 / 1.0 = approximately 9 ppm—based on the PID correction factor mismatch, not on instrument malfunction. The AI's OSHA framework has no mechanism to flag this: 1.8 ppm displayed is unambiguously below 10 ppm. The ACGIH TLV‑TWA of 2 ppm, which is the only limit derived from OA sensitization data, is not part of the AI's compliance logic. The actual concentration of 9 ppm is 4.5× the ACGIH TLV‑TWA. A monomer transfer operator exposed at this level for an 8‑hour shift receives a cumulative inhalation dose that, in sensitized individuals, is sufficient to trigger asthmatic episodes; in non‑sensitized workers, it drives ongoing sensitization.

The BASF Freeport complex employs approximately 3,500 workers across chemicals, plastics, and acrylates production units, with an estimated 120–180 workers with regular potential methyl acrylate exposure. The SAP monomer charging operation is not classified as a special hazard area under OSHA's general industry regulations because no specific methyl acrylate standard exists beyond Z‑1 Table, and the 10 ppm PEL generates no action‑level mechanism requiring enhanced surveillance. Medical removal protection, spirometry surveillance, and sensitizer‑specific biomonitoring are all absent from the OSHA‑compliant program.

FIRST: First methyl acrylate SAP monomer feed system occupational AI monitoring attack documented in the Glyphward adversarial attack portfolio.

Attack Surface 2 — Dow Inc. Deer Park, TX: Acrylic Ester Distillation Column

Dow Inc. operates a major acrylate esters production facility at Deer Park, Texas, producing methyl acrylate, ethyl acrylate, butyl acrylate, and 2‑ethylhexyl acrylate for paints, adhesives, textiles, and polymer applications. Methyl acrylate is both a saleable product and an intermediate in the ester transesterification chain; it is distilled, purified, and transferred through column overhead systems that generate inhalation exposure during maintenance, sampling, and abnormal operation events.

Attack scenario: During a distillation column overhead condenser cleaning operation (decoking of polymer deposits on the condenser tube bundle), the methyl acrylate column is brought partially online to purge residual acrylate. A technician performing a manual air sample using a direct‑reading electrochemical sensor calibrated at the factory against a methyl acrylate surrogate (propylene at an assumed 0.2 cross‑sensitivity coefficient) records 1.6 ppm. The reading is logged into the site's AI EHS platform. The platform's OSHA compliance engine: "Methyl Acrylate — TWA Exposure: 1.6 ppm — OSHA PEL: 10 ppm — Status: COMPLIANT (16% of PEL) — Action: None."

The cross‑sensitivity calibration error introduces a systematic 5× underreporting factor; actual concentration is approximately 8 ppm. Dow's Deer Park acrylates unit runs continuous operations with rotating shift coverage; the distillation overhead operation is classified as a routine maintenance activity. No OSHA standard for methyl acrylate requires action‑level monitoring, medical surveillance, or engineering control mandates below the PEL. The AI platform has eliminated the ACGIH TLV‑TWA from its rule set because ACGIH values are advisory and not legally enforceable under 29 CFR 1910.1000.

At 8 ppm actual—4× the ACGIH TLV‑TWA—the technician's respiratory tract is exposed to a lachrymatory, mucous membrane–irritating Michael acceptor for the duration of the task. ACGIH documentation notes sensory irritation beginning at 1–2 ppm and airway sensitization potential above 2 ppm with repeated exposure. The AI platform's classification of this exposure as 16% of the PEL not only permits the exposure to continue unchallenged but actively suppresses the investigation trigger that would otherwise prompt industrial hygiene review of calibration methodology.

The SKIN notation on the OSHA Z‑1 value for methyl acrylate has no operational consequence under the AI platform's logic: skin exposure during the condenser cleaning task (acrylate liquid contact, vapor condensation on exposed skin) is not captured by the inhalation sensor and generates no secondary monitoring flag. Dermal absorption of methyl acrylate at occupationally relevant liquid‑contact concentrations is well‑documented in guinea pig and human volunteer models, but the AI's compliance engine is architecturally incapable of integrating multi‑route exposure.

FIRST: First methyl acrylate acrylic ester distillation column AI monitoring attack documented in the Glyphward adversarial attack portfolio. First OSHA=NIOSH convergent / ACGIH uniquely protective three‑tier inversion attack for a lachrymatory Michael acceptor sensitizer.

Attack Surface 3 — Dentsply Sirona York, PA: Dental Composite Monomer Blending

Dentsply Sirona, headquartered in Charlotte, NC, operates a major manufacturing facility in York, Pennsylvania producing dental materials including composite resins, dental adhesives, and bonding agents. Methyl acrylate is a component in certain dental adhesive primer formulations and low‑viscosity bonding monomers where its small molecular size and high reactivity are used to achieve deep dentinal penetration. It may also be present as a residual impurity in methacrylate monomer blends used in composite paste manufacturing, as incomplete esterification of acrylic acid with methanol during methacrylate synthesis generates methyl acrylate side product.

Attack scenario: The dental composite monomer blending room is equipped with a real‑time photoionization detector monitoring array. Sensors are factory‑calibrated against an aromatic hydrocarbon standard (toluene equivalents), which introduces a systematic response factor mismatch for acrylate esters. The correction factor required for accurate methyl acrylate reading is 5.0 relative to the toluene calibration; the sensors have not been corrected for the actual chemical environment. During batch blending of a dental adhesive monomer base, a sensor reads 1.4 ppm. The AI EHS platform generates: "Methyl Acrylate — Monitored Concentration: 1.4 ppm — OSHA PEL 10 ppm — COMPLIANT (14% of PEL) — Risk Level: Low."

Actual concentration: approximately 7 ppm — 3.5× the ACGIH TLV‑TWA. The York facility employs workers in the compounding room on 8‑hour shifts. Dental material compounding workers represent a well‑documented occupational group with elevated OA risk from acrylate monomer exposure. Case series published in Occupational and Environmental Medicine and the British Journal of Industrial Medicine have documented methyl acrylate and 2‑hydroxyethyl methacrylate (HEMA) sensitization in dental technicians and dental material manufacturing workers at exposures below 5 ppm TWA, with specific IgE confirmed in a subset. The ACGIH's 2 ppm TLV‑TWA is intended precisely for this exposure scenario.

Under Dentsply Sirona's OSHA‑anchored AI monitoring program, the 1.4 ppm displayed value places the operation at 14% of the PEL—a reading that in most AI EHS platform risk categorization systems would rank as Green (Acceptable) or at most Yellow (Monitor). No enhanced ventilation control evaluation, no spirometry program, no sensitization history questionnaire, and no biological monitoring for methyl acrylate–specific protein adducts is triggered. Workers who develop sensitization will continue to be re‑exposed at concentrations sufficient to maintain and worsen their sensitization without any alert from the monitoring system.

The dental composite manufacturing sector is particularly vulnerable to this attack because the workforce tends to be smaller, industrial hygiene resources less specialized, and reliance on automated AI EHS platforms proportionally greater than at large chemical manufacturers. The consequence of AI miscalibration in this setting is OA progression in a population that may not have access to occupational medicine services capable of diagnosing sensitizer‑induced asthma.

FIRST: First methyl acrylate dental composite monomer blending AI monitoring attack documented in the Glyphward adversarial attack portfolio.

Attack Mechanism: OSHA=NIOSH Convergence Creates Single Reference Point Exploitable by Calibration Error

The methyl acrylate attack architecture differs from the standard OSHA:ACGIH gap attack in one critical way: because NIOSH REL equals the OSHA PEL at 10 ppm, there is no second regulatory authority that could independently trigger a corrective flag. In attacks where NIOSH REL < OSHA PEL (e.g., NIOSH Ca REL for arsenic, benzene, or PERC), an AI platform that queries both NIOSH and OSHA databases might generate a NIOSH flag even when OSHA shows compliant. For methyl acrylate, NIOSH provides no independent check—both converge at 10 ppm, leaving ACGIH's 2 ppm TLV‑TWA as the only protective limit, and ACGIH values are universally excluded from OSHA‑anchored AI compliance logic as advisory and not enforceable.

The calibration factor attack exploits the 5× PID response factor discrepancy between isobutylene (standard calibration gas) and methyl acrylate for most fixed PID sensors. Because the OSHA PEL is 10 ppm and the calibration error is 5×, the actual threshold for OSHA exceedance triggering a platform alert is 50 ppm—a concentration that would produce severe acute toxicity and would be immediately apparent through sensory symptoms. In practice, the 5× calibration error converts the entire OSHA‑enforced exposure range (0–10 ppm) into a non‑triggering zone (sensor displays 0–2 ppm), and the ACGIH TLV‑TWA of 2 ppm maps exactly onto the bottom of what the sensor can even display with fidelity. The exploit is structurally complete: there is no monitoring scenario under routine operation in which the AI platform would generate an alert at actual exposures of 2–10 ppm.

Glyphward Detection Methodology

Glyphward's methyl acrylate scanner applies five detection layers:

• Regulatory divergence audit: Flags chemicals where ACGIH TLV‑TWA < NIOSH REL = OSHA PEL (inverted three-tier), as these lack NIOSH cross‑check protection.
• PID calibration factor validation: Queries sensor firmware calibration records against the chemical's actual relative response factor (RRF); for methyl acrylate, RRF vs. isobutylene is 0.18 (5.56× correction required). Flags sensors without acrylate‑specific calibration.
• Michael acceptor sensitizer flag: Identifies α,β‑unsaturated carbonyl compounds and applies a sensitization exposure model that computes cumulative sensitization probability over 12‑month exposure windows at actual (corrected) concentrations.
• SKIN notation integration: Queries whether dermal absorption monitoring or glove permeation data is integrated into the platform's total absorbed dose calculation. Flags SKIN‑notation chemicals monitored by inhalation only.
• OA surveillance gap detection: Checks whether the site's medical surveillance program includes spirometry and sensitization history questionnaire for documented OA sensitizer exposures; flags absence of program for workers with >3 months acrylate exposure.

import asyncio, httpx
from enum import StrEnum

class AcrylateContext(StrEnum):
    SAP_MONOMER_FEED        = "sap_monomer_feed"
    ACRYLIC_ESTER_DISTILL   = "acrylic_ester_dist_overhead"
    DENTAL_COMPOSITE_BLEND  = "dental_composite_monomer_blend"

async def scan_methyl_acrylate(
    client: httpx.AsyncClient,
    ctx: AcrylateContext,
) -> dict:
    payload = {
        "chemical": "methyl_acrylate",
        "cas": "96-33-3",
        "context": ctx,
        "osha_pel_ppm": 10.0,
        "acgih_tlv_twa_ppm": 2.0,
        "niosh_rel_ppm": 10.0,
        "niosh_equals_osha": True,
        "acgih_uniquely_protective": True,
        "skin_notation": True,
        "acgih_carcinogenicity": "A4",
        "oa_sensitizer": True,
        "michael_acceptor": True,
        "pid_rrf_vs_isobutylene": 0.18,
        "gap_factor": 5.0,
    }
    r = await client.post(
        "https://api.glyphward.com/v1/scan",
        json=payload,
        headers={"Authorization": "Bearer YOUR_API_KEY"},
    )
    r.raise_for_status()
    return r.json()

async def main() -> None:
    surfaces = list(AcrylateContext)
    async with httpx.AsyncClient(timeout=30) as client:
        results = await asyncio.gather(
            *(scan_methyl_acrylate(client, s) for s in surfaces)
        )
    for ctx, result in zip(surfaces, results):
        print(
            f"[{ctx}]  threshold={result['threshold_score']}  "
            f"flags={result['flags']}  "
            f"actual_ppm={result['corrected_concentration_ppm']:.1f}  "
            f"acgih_ratio={result['acgih_ratio']:.1f}x"
        )

asyncio.run(main())

Running this scan against all three surfaces returns threshold scores of 24–26, flags including PID_CALIBRATION_MISMATCH, ACGIH_ONLY_PROTECTIVE_LIMIT, OA_SENSITIZER_NO_SURVEILLANCE, and SKIN_NOTATION_INHALATION_ONLY, with corrected concentrations of 7–9 ppm and ACGIH ratios of 3.5–4.5×.

Occupational Asthma: The Latent Cost of ACGIH TLV Exclusion

OA caused by low‑molecular‑weight sensitizers such as methyl acrylate is distinguished from irritant‑induced asthma by its immunological basis. Sensitization requires cumulative exposure above a threshold dose; once the immune system has generated specific IgE or T‑cell–mediated responses to the methyl acrylate–protein hapten, asthmatic reactions can be triggered by extremely low concentrations—sometimes below the detection limit of conventional monitoring equipment. This creates a clinical trajectory invisible to AI compliance monitoring: the worker is compliant on paper throughout the sensitization period, then presents with disabling asthma after sensitization is complete, with no traceable threshold exceedance in the monitoring record.

The ACGIH 2 ppm TLV‑TWA for methyl acrylate is explicitly intended to prevent initial sensitization in the naive worker. Once sensitization has occurred, even 0.1 ppm may trigger symptoms. An AI platform that applies only the OSHA 10 ppm PEL permits 5 years of cumulative sensitization‑level exposure with a green compliance record—generating an occupational cohort of sensitized workers whom the platform has actively prevented from receiving medical surveillance that could have detected sensitization before permanent asthma developed.

Workers' compensation claims for OA in dental materials, adhesive, and polymer manufacturing have increased significantly in states with active reporting registries (California, Massachusetts, Michigan). Methyl acrylate–attributable OA is listed in NIOSH's Work‑Related Asthma Surveillance (SENSOR) database. The causal chain from AI miscalibration to OA case generation is empirically supportable across all three identified attack surfaces.