Low-Fidelity Biology, Entropic Waste, and the End of the One Disease/One Cause Safety Model

An RF Safe educational manifesto for moving wireless safety beyond heat accounting and toward biological signal integrity

The same field is not the same biological event in every body.
The receiver is part of the dose.
The absence of a one-to-one disease map is not evidence of biological inertness.

RF safety has been asking the wrong question

For decades, the public debate has been organized around one deceptively simple question:

Does radiofrequency radiation cause Disease X?

Does it cause glioma? Does it cause infertility? Does it cause a particular neurological disorder? Does it cause a specific immune condition?

These are legitimate questions, but they begin at the far end of the biological chain. A named disease is usually the visible endpoint of a long process involving susceptibility, timing, compensatory capacity, developmental stage, co-exposures, repair, and chance. By demanding that every exposure map cleanly onto a single diagnosis, the conventional framework may overlook what is happening upstream—where living systems first detect, interpret, compensate for, or fail to compensate for an environmental disturbance.

The more fundamental question is:

What does a structured electromagnetic field do to the timing, thresholds, phase relationships, and recovery dynamics of biological control systems—and in which biological receiver?

That question changes everything.

The categorical claim that radiofrequency electromagnetic fields below established heating-injury thresholds are biologically inert is no longer defensible. ICNIRP’s own 2020 guidelines acknowledge that RF exposure at specific absorption rates below 2 W/kg can affect waking alpha activity and the sleep-spindle range of the EEG, even though ICNIRP does not classify those observations as established adverse health effects. A 2025 double-blind human experiment found that a controlled 3.6 GHz exposure modified sleep-spindle frequency in a genotype-dependent manner. FDA has also authorized a medical device that uses very-low-level, specifically amplitude-modulated RF fields to produce a probable therapeutic benefit in advanced liver cancer. These findings do not establish that every wireless signal is harmful. They establish the narrower—and scientifically decisive—point that temperature is not the only biologically meaningful variable.

That distinction is the foundation of the RF Safe perspective:

Biological activity below a heating threshold is established. The unresolved questions concern adversity, susceptibility, waveform, timing, cumulative exposure, and long-term consequence.

Biology is not merely chemistry. It is timed chemistry.

A living organism is not simply a container filled with molecules. It is a coordinated, adaptive control system.

Life depends on signals being delivered:

• at the correct time;
• in the correct sequence;
• at the correct amplitude;
• in the correct cellular compartment;
• for the correct duration;
• at the correct developmental stage;
• followed by the correct recovery or termination signal.

Calcium illustrates this principle exceptionally well. Calcium is not merely a substance that rises or falls. Its oscillation frequency carries information. Classic experimental work demonstrated that changing calcium-oscillation frequency can change the efficiency and specificity with which transcription factors and genes are activated. Rapid and infrequent calcium pulses do not necessarily convey the same instruction, even when a crude average of calcium concentration appears similar.

This means that biological dose cannot always be reduced to total energy or average concentration. Timing is part of the dose. Pattern is part of the dose. Phase is part of the dose.

A safety model centered primarily on average energy absorption can therefore miss a critical category of interaction: a disturbance that is too weak to injure tissue thermally but sufficiently structured to alter biological timing.

This is not an exotic proposition. Medicine already uses timed electrical, magnetic, acoustic, optical, and RF inputs to influence biology. The response to such inputs depends on their pattern and on the state of the receiving tissue. A field may be therapeutically useful in one waveform, physiologically detectable in another, and irrelevant in a third.

Energy matters—but energy is not the whole message.

Biological fidelity

RF Safe uses biological fidelity to mean the precision with which a living system preserves and executes biological information across time.

High-fidelity biology maintains coherent relationships among membrane voltage, ion-channel gating, calcium oscillations, mitochondrial metabolism, redox signaling, gene expression, neural rhythms, immune responses, repair, apoptosis, and tissue organization.

A high-fidelity system does not merely contain the correct molecular parts. It deploys those parts in the correct place, sequence, rhythm, and proportion.

A radio receiver may contain every required component yet fail to reproduce a message when noise, timing error, phase distortion, or poor synchronization enters the system. Biology is vastly more complex, but the systems principle remains relevant: having the correct components does not guarantee a faithful output when control signals lose precision.

Low-fidelity biology

Low-fidelity biology is RF Safe’s term for a state in which biological regulation becomes less precise, more variable, and less coherent.

It is not itself a conventional diagnosis. It is an upstream systems condition that may involve:

• increased timing jitter in calcium or neural signaling;
• altered ion-channel activation thresholds;
• inappropriate signaling in the wrong compartment or developmental window;
• transcriptional noise;
• disturbed mitochondrial-redox coupling;
• maladaptive immune set-point changes;
• impaired repair or clearance;
• accumulated compensatory strain;
• reduced resilience to subsequent environmental challenges.

The same decline in regulatory fidelity need not produce the same disease in every person. One individual’s weak point may be neurological. Another’s may be immunological, reproductive, cardiovascular, metabolic, or developmental. A third person may compensate successfully for decades.

This is why demanding a single RF-to-single-disease relationship may be conceptually inadequate. A general upstream disturbance can produce multiple downstream outcomes, while a single downstream diagnosis can arise from multiple upstream disturbances.

Low-fidelity biology reframes the inquiry from:

“Which disease does this exposure cause?”

to:

“Which control processes does this exposure disturb, under which conditions, in which receiver, and with what cumulative loss of resilience?”

Entropic waste

RF Safe uses entropic waste as a systems-biology term for the accumulated burden created when biological information and energy are repeatedly processed with declining fidelity.

Entropic waste is not a pathogen, a literal infectious agent, or a single chemical substance. Nor is the term intended to replace the precise thermodynamic definition of entropy.

It describes the biological consequences of unresolved error:

• mistimed signals;
• damaged or poorly cleared cellular components;
• inappropriate proteins or signaling states;
• redox imbalance;
• mitochondrial inefficiency;
• repair debt;
• maladaptive compensation;
• chronic inflammatory signaling;
• lost synchronization among regulatory networks.

In engineering language, a noisy system expends additional energy identifying, correcting, bypassing, or compensating for errors. When the error burden exceeds the system’s correction capacity, fidelity declines further. RF Safe proposes that a comparable principle should be investigated in biology: repeated environmental perturbation may create biological costs even when no single exposure produces an immediate, diagnosable injury.

The scientific task is to convert this concept into measurable endpoints. Entropic waste should eventually be evaluated through operational markers such as signaling variability, calcium-waveform distortion, redox burden, mitochondrial compensation, protein-clearance deficits, transcriptional noise, immune-state drift, and reduced recovery after controlled physiological challenges.

A term becomes scientifically useful not because it sounds powerful, but because it generates falsifiable predictions.

The meta-disease state

A meta-disease state is the proposed downstream consequence of persistent low-fidelity biology: a prediagnostic loss of coherence and resilience that can increase susceptibility to several conventional diseases without being identical to any one of them.

This does not mean that all illnesses have the same cause. It means that many illnesses may share upstream vulnerabilities: impaired regulation, chronic oxidative or inflammatory burden, poor repair, altered bioelectric signaling, disrupted developmental timing, or reduced adaptive capacity.

The meta-disease model therefore does not require RF exposure to cause every downstream condition directly. It asks whether RF can act as one contributor to an upstream state in which the body becomes less capable of maintaining order when challenged by other biological, chemical, physical, infectious, nutritional, or psychosocial stressors.

In this model, the central outcome is not always a tumor, symptom, or diagnosis. It may first be a measurable loss of regulatory precision.

The receiver is part of the dose

Traditional exposure science emphasizes the source: frequency, field strength, power density, and specific absorption rate.

Those variables are essential, but they do not completely describe the biological event. The receiver also matters.

RF Safe expresses the relationship conceptually as:

Biological response = f (field architecture × receiver architecture × current biological state × exposure history × co-exposures)

Field architecture includes carrier frequency, modulation, pulse structure, peak intensity, average intensity, duty cycle, polarization, duration, repetition, spatial distribution, and near-field or far-field conditions.

Receiver architecture includes genotype, epigenetic state, ion-channel expression, membrane composition, tissue geometry, mitochondrial state, age, sex, developmental stage, and the local density of biologically responsive structures.

Current biological state includes membrane voltage, calcium state, redox balance, circadian phase, hormonal context, immune activation, sleep stage, temperature, hydration, metabolism, and stress physiology.

Exposure history includes cumulative exposure, adaptation, sensitization, repair debt, prior injury, and the interval between exposures.

Co-exposures include other physical, chemical, nutritional, pharmaceutical, infectious, and environmental inputs acting on the same regulatory systems.

This is not presented as a completed quantitative law. It is a causal map for designing better experiments.

Its central proposition is straightforward:

The same external field can become a different biological event when it passes through a different receiver.

The CACNA1C sleep-spindle study: one input, two receivers

The 2025 NeuroImage study involving the CACNA1C variant rs7304986 provides a compelling human example of receiver-dependent biology.

Thirty-four genotyped participants—15 T/C carriers and 19 matched T/T carriers—underwent randomized, double-blind, sham-controlled sessions involving 30 minutes of pre-sleep exposure to 700 MHz, 3.6 GHz, or sham conditions. Sleep activity was then measured using high-density EEG.

The significant effect was not uniform. After the 3.6 GHz exposure, T/C carriers exhibited an increase in NREM sleep-spindle center frequency across central, parietal, and occipital regions. Matched T/T carriers did not show the same response. The investigators concluded that the 3.6 GHz signal modulated sleep-spindle frequency in a CACNA1C genotype-dependent manner and implicated L-type calcium-channel physiology in the response.

This was a small study, and it did not establish neurological injury or future disease. Its importance lies elsewhere.

It directly challenges the assumption of a uniform biological receiver.

The exposure was standardized. The measured outcome was objective. The response depended on genotype.

The rs7304986 variant is located in a noncoding intronic region of CACNA1C, the gene encoding the α1C subunit of the Cav1.2 L-type voltage-gated calcium channel. RF Safe’s earlier ceLLM analysis interprets this result as evidence that noncoding genomic architecture can influence how an external electromagnetic input is translated into a measurable bioelectric rhythm.

The finding can be summarized in three lines:

One letter changed the receiver.
The receiver changed the rhythm.
The rhythm revealed the logic layer.

The result does not mean that CACNA1C is a “5G gene.” CACNA1C is part of native human bioelectric machinery. Cav1.2 channels already participate in converting membrane-voltage changes into calcium signals involved in neural activity, gene regulation, cardiovascular function, development, and sleep physiology.

The external field did not create that machinery. It served as a probe that revealed a receiver difference already present in the biology.

That is the revolutionary point: susceptibility is not merely noise around an average response. Susceptibility is part of the mechanism.

ceLLM: from genetic blueprint to weighted biological receiver

RF Safe’s Cellular Latent Learning Model, or ceLLM, proposes that the cell should not be understood solely as a machine executing a flat genetic program.

In the ceLLM framework, the genome, chromatin topology, membrane potential, calcium dynamics, redox state, mitochondrial activity, tissue context, and environmental signals form a continuously weighted biological network. DNA provides components, but the current physical and regulatory state of the cell influences how those components are queried and deployed.

The analogy to a weighted information system is useful:

• the DNA sequence contributes the available architecture;
• chromatin state and three-dimensional folding influence accessibility and regulatory weighting;
• membrane voltage and calcium act as state-dependent inputs;
• mitochondrial and redox systems constrain available energy and signaling;
• developmental history and epigenetics alter response thresholds;
• external fields and other environmental inputs become additional perturbations or queries.

ceLLM is not yet an established biological theory. It is a falsifiable research program. It predicts that electromagnetic responses will often be conditional rather than uniform; that noncoding variation can affect field sensitivity; that the same waveform may produce different outputs at different developmental or circadian stages; and that multi-omics changes may precede recognizable disease.

The CACNA1C sleep-spindle study is important because its architecture resembles that prediction:

same external input, different internal weighting, different bioelectric output.

S4 helices: biology already contains electric-field sensors

Voltage-gated sodium, potassium, and calcium channels contain voltage-sensing domains composed of four transmembrane helices, S1 through S4. The S4 helix contains regularly spaced positively charged residues that act as gating charges. These charges directly sense changes in the transmembrane electric field, and S4 movement helps control whether the channel opens or closes.

This is settled biophysics: living cells contain molecular structures whose function is to detect electrical conditions.

That fact does not, by itself, prove that ambient Wi-Fi, cellular, or 5G signals disrupt S4 movement. The coupling problem must be demonstrated under realistic exposure conditions. It requires appropriate dosimetry, thermal control, waveform characterization, electrophysiology, replication, and tissue-specific analysis.

But S4 architecture makes the mechanistic question legitimate:

Can a non-native, time-varying field—or an induced secondary electrical or biochemical process—alter the probability, timing, or synchrony of voltage-sensor gating in a susceptible biological state?

That is a scientific question, not a metaphysical one.

A small shift in channel-opening probability may be biologically insignificant in one context. In another context—during neural synchronization, cardiac conduction, immune activation, embryonic patterning, or calcium-dependent transcription—the timing of the shift may matter more than its average magnitude.

The correct experimental endpoint is therefore not always cell death or temperature. It may be:

• gating kinetics;
• activation threshold;
• open-state probability;
• refractory timing;
• calcium-oscillation frequency;
• phase synchronization;
• downstream gene expression;
• recovery after exposure.

CYB5B: a demonstrated EMF-to-calcium-to-gene pathway

A 2026 study in Cell developed an electromagnetic-field-inducible gene switch for remotely controlling gene expression. Using a CRISPR-Cas9 screen, the investigators identified cytochrome b5 type B—Cyb5b—as an essential mediator that likely acted as an EMF sensor in their engineered system.

Most importantly, activation depended on rhythmic calcium oscillations rather than generic bulk calcium influx. The field did not simply “add calcium.” It produced a specific calcium-dynamic pattern that activated the gene switch.

This paper used a defined, purpose-built low-frequency EMF system; it was not a study of ordinary Wi-Fi, cellular, or 5G exposure. It therefore does not establish that CYB5B mediates the biological effects of telecommunications RF.

What it does establish is still profound:

1. A controlled electromagnetic field can be coupled to a specific molecular mediator.
2. That mediator can generate rhythmic calcium dynamics.
3. Calcium timing can control gene expression.
4. The effect can be used deliberately in living systems.

CYB5B therefore belongs in the RF Safe research agenda as a candidate mechanistic lever—not as a prematurely declared universal microwave sensor.

The next experiments are clear: determine whether CYB5B dependence appears under RF waveforms, intensities, modulation patterns, and biological conditions relevant to real-world wireless exposure. Knockout, rescue, calcium imaging, electrophysiology, transcriptomics, redox measurements, and exact thermal control should be used.

A revolutionary model does not merely announce a mechanism. It specifies the experiment that could disprove it.

Density gating

RF Safe uses density gating to describe the hypothesis that biological response depends partly on the local number, distribution, geometry, and state of susceptible targets.

A tissue containing a high density of voltage-gated channels, tightly synchronized excitable cells, specialized mitochondrial organization, or a particular CYB5B/calcium-signaling architecture may respond differently from a tissue lacking those conditions.

Density gating could help explain:

• tissue-specific responses;
• narrow frequency or modulation windows;
• non-monotonic dose-response relationships;
• developmental sensitivity;
• effects appearing only during certain sleep or circadian phases;
• genotype-specific responses;
• failure to replicate an effect when the relevant biological state is absent.

This remains an RF Safe hypothesis, but it is testable. Target density can be quantified. Tissue architecture can be mapped. Expression levels can be altered. Responsive and nonresponsive tissues can be compared under identical exposure conditions.

The broader principle is that exposure magnitude alone may not determine response; the organization of the receiving system may set the gate.

A field can be medicine, probe, or noise

The TheraBionic P1 device is an instructive example of why RF science cannot be reduced to “ionizing versus non-ionizing” or “hot versus not hot.”

The device uses very-low-level RF electromagnetic fields derived from amplitude modulation of a 27.12 MHz carrier. FDA granted it a Humanitarian Device Exemption for adults with advanced hepatocellular carcinoma after failure of first- and second-line therapy, concluding that the available data supported reasonable assurance of safety and probable benefit for that limited population.

The FDA documentation describes tumor-specific modulation frequencies and identifies calcium-channel blockers as a treatment concern. It also cites research involving Cav3.2 T-type voltage-gated calcium channels and calcium influx.

A Humanitarian Device Exemption is not the same as conventional proof of effectiveness from large randomized trials. The approval rests on probable benefit in a small, severely ill population.

Nevertheless, the device demonstrates a principle that should permanently end simplistic claims of universal RF inertness:

A weak, structured RF signal can be designed to produce a biologically meaningful response.

That does not mean ordinary wireless signals produce the same response. It means frequency, modulation, timing, receiver state, and biological target matter.

A field can be medicine when its pattern is selected for a therapeutic target.

It can be a probe when it reveals a genotype-dependent sleep rhythm.

It can become biological noise when an imposed signal interferes with native timing without producing a beneficial or adaptive output.

The relevant question is not simply how much energy was absorbed.

The relevant question is: What information-bearing pattern reached what receiver at what time?

The animal evidence: a convergence that cannot be dismissed

The U.S. National Toxicology Program conducted large, long-term rodent studies using 900 MHz GSM- and CDMA-modulated signals in rats and 1900 MHz signals in mice.

NTP concluded that there was clear evidence of an association between exposure and malignant heart schwannomas in male rats, some evidence involving malignant gliomas in the brains of male rats, and some evidence involving adrenal pheochromocytomas. NTP also reported DNA damage in selected tissues in follow-up analyses. Results in female rats and in mice were less clear.

The Ramazzini Institute then reported results from a lifetime, far-field-style 1.8 GHz GSM exposure study. Its findings included increased heart schwannomas in male rats at the highest exposure and glial-cell lesions that contributed to concern because the tumor types overlapped with the NTP signal despite major differences in exposure design and intensity.

A 2025 systematic review published in Environment International—partially funded by WHO and included in the WHO RF-EMF systematic-review series—evaluated 52 animal studies, including 20 chronic bioassays. It found no or minimal evidence of exposure-related cancer in most organ systems. However, it rated the certainty of evidence as high for increased gliomas and malignant heart schwannomas in male rats. It also emphasized that translating these findings to human risk is complex because mechanism, dose metric, localized versus whole-body exposure, cumulative exposure, and the shape of the dose-response relationship remain unresolved.

That wording matters.

The review did not declare that wireless exposure causes human cancer. It did not find cancer increases in every organ. It found high-certainty animal evidence for two specific tumor outcomes in male rats.

That is not proof of universal human harm, but neither is it compatible with the claim that the animal cancer evidence is nonexistent.

The tumors were not merely labels on a pathology sheet

A 2024 PLOS ONE study examined genetic alterations in gliomas and cardiac schwannomas arising in rats from the Ramazzini lifetime RF study.

The investigators reported that the rat gliomas histologically resembled low-grade human gliomas. They did not find homologues of the classic human IDH1 or IDH2 hotspot mutations, but they found that the tumors appeared to share some alterations with IDH-wildtype human gliomas. Cardiac schwannomas also carried mutations in some of the queried cancer-related genes.

The responsible conclusion is not that the rat and human tumors were genetically identical. They were not.

The significant conclusion is that the tumors displayed partial histological and molecular features relevant to human disease. That strengthens the translational question and supports further comparative work rather than dismissal of the animal findings as biologically meaningless.

The negative animal studies must be included

A credible manifesto cannot omit evidence that complicates its preferred interpretation.

In 2026, Japanese and Korean teams published coordinated long-term studies designed to test aspects of the NTP findings. Male rats were exposed to 900 MHz CDMA-modulated RF at a whole-body SAR of 4 W/kg. Neither study found statistically significant increases in brain, heart, or adrenal tumors, and neither found evidence of genotoxicity under its protocol.

These studies matter. They failed to reproduce the NTP tumor signal at the single exposure level they tested.

They do not erase NTP, Ramazzini, or the 2025 systematic review. Nor can they establish that every RF waveform and exposure condition is safe. The Japanese and Korean studies used one SAR level, one modulation system, male rats, and a defined protocol; German federal reviewers noted that their statistical power could not rule out every effect on rare heart or brain tumors.

The combined animal record therefore rejects two oversimplifications:

• It does not support the claim that RF exposure invariably produces tumors.
• It does not support the claim that the positive animal tumor findings never occurred or have no scientific significance.

The evidence is signal-, protocol-, species-, sex-, dose-, and state-dependent.

That complexity is precisely why a one-number safety model is inadequate.

Human cancer epidemiology constrains the claim—but does not establish inertness

A WHO-commissioned systematic review of human observational studies reported moderate-certainty evidence that near-field exposure from mobile-phone use likely does not increase the risk of several major brain and head tumors in adults, including glioma, meningioma, acoustic neuroma, pituitary tumors, and salivary-gland tumors. It also found moderate-certainty evidence that far-field exposure from fixed-site transmitters likely does not increase childhood leukemia risk, while evidence for several other exposure-outcome combinations was limited or unavailable.

These findings should not be ignored. They argue against a large, uniform cancer effect that would be readily detectable across the general population using the exposure proxies and time periods represented in the available studies.

But they do not demonstrate biological inertness.

Epidemiology organized around subscription history, self-reported phone use, estimated call time, or residential proximity may be poorly equipped to detect:

• modulation-specific effects;
• genotype-defined susceptibility;
• developmental windows;
• sleep-stage interactions;
• peak or pulse effects obscured by averaging;
• multiple simultaneous RF sources;
• rapidly changing technologies;
• long latency extending beyond available follow-up;
• upstream physiological changes that never become cancer.

Human epidemiology remains indispensable. The lesson is not to discard it. The lesson is to improve exposure classification and stop asking it to answer mechanistic questions its datasets were not designed to resolve.

Why Disease X is difficult to map

A clean exposure-to-disease relationship is easiest to detect when several conditions hold: exposure is accurately measured, biological response is relatively uniform, the dose-response relationship is monotonic, the latency is known, the endpoint is specific, and competing causes are limited.

RF exposure rarely satisfies those conditions.

Two people identified as “phone users” may have radically different exposures because of device position, transmit power, network quality, antenna configuration, call duration, data use, Wi-Fi use, Bluetooth use, shielding by buildings, and distance from infrastructure.

Even accurately measured external fields do not guarantee equal internal response. Genotype, tissue geometry, ion-channel expression, age, sleep state, hormones, redox balance, and prior exposure may change the biological event.

Averaging can further obscure structure. Two signals with the same average power can have different peaks, duty cycles, pulse trains, and modulation envelopes. A thermal metric may treat them as similar while a timing-sensitive biological system does not.

Finally, a loss of fidelity may not immediately become a disease. Compensation can conceal upstream disturbance for years. A susceptible individual may cross a pathological threshold while another does not. One person may develop an immune phenotype; another may experience sleep or neurological disruption; another may remain clinically unaffected.

The inability to map every exposure directly to Disease X is therefore not proof that nothing happened upstream.

It may be evidence that the original causal model was too small.

What the Melnick–Moskowitz risk analysis actually found

In March 2026, Ronald L. Melnick and Joel M. Moskowitz published an analysis applying benchmark-dose and conventional health-risk assessment methods to animal cancer and reproductive data.

Using EPA Benchmark Dose Tools and linear low-dose extrapolation under stated assumptions, the authors estimated whole-body SAR values of approximately 0.8 to 5 mW/kg for an additional cancer risk of one in 100,000, depending partly on daily exposure duration. For male reproductive protection, they derived values of approximately 3.3 to 10 mW/kg.

They compared those estimates with the existing public whole-body limit of 80 mW/kg. Their conclusion was that current limits are approximately 15 to 900 times higher than their cancer-risk-based estimates and 8 to 24 times higher than their estimates for protecting male reproductive health.

The commonly repeated “900 times for cancer and 24 times for fertility” language refers to the upper ends of those ranges.

This was not an EPA determination, and the resulting numbers are not official government exposure limits. They are the authors’ application of EPA-style risk-assessment methods, including assumptions about low-dose extrapolation, acceptable risk, exposure duration, and the relevance of animal evidence to humans.

But that qualification does not make the analysis unimportant.

It exposes a regulatory question that must be answered directly:

Why are RF limits not routinely tested against the same benchmark-dose, uncertainty-factor, susceptible-population, and lifetime-risk methodologies used for chemical toxicants and other potential carcinogens?

Disagreement with the authors’ assumptions should lead to an open competing analysis—not silence.

The FCC lost because it failed to explain itself

In 2021, the U.S. Court of Appeals for the D.C. Circuit remanded the FCC’s decision to retain its 1996 exposure framework.

The court did not rule that RF exposure causes cancer or declare the FCC limits scientifically unsafe. It ruled that the FCC failed to provide the reasoned explanation required by administrative law.

The court specifically identified failures involving non-cancer health evidence, effects on children, long-term exposure, pulsing and modulation, the ubiquity of wireless technology, technological developments since 1996, portable-device testing procedures, and environmental effects.

The FCC therefore did lose.

It lost not because judges resolved the scientific debate, but because the agency could not lawfully close the record while failing to engage with central questions raised by that record.

That is a critical distinction—and an indictment of the regulatory process in its own right.

A standard does not become scientifically adequate merely because devices comply with it.

Compliance proves conformity to the standard.

It does not prove that the standard measures every biologically relevant variable.

The FDA’s public assurances are no longer a stable foundation

In January 2026, HHS announced a new study intended to identify research gaps concerning electromagnetic radiation and modern technologies. At the same time, FDA removed older webpages containing categorical cell-phone safety conclusions that HHS described as outdated.

That action was not a formal FDA admission that cell phones cause disease. FDA pages carrying broad safety assurances have also remained accessible, leaving the federal government’s public messaging internally inconsistent.

The strongest accurate formulation is:

HHS reopened the question, and FDA withdrew some of its older categorical messaging.

That is a walk-back from presenting the matter as permanently settled. It is not a confession of harm.

For public policy, the implication is substantial. Agencies cannot rely indefinitely on old assurances while initiating new reviews because the old evidentiary base no longer resolves the modern exposure environment.

The “thermal-only” dispute needs precise language

Calling the ICNIRP framework “thermal-only” captures an important criticism, but the phrase should be used carefully.

ICNIRP states that it evaluated literature described as low-level or nonthermal. It also acknowledges reproducible EEG changes below 2 W/kg. However, ICNIRP does not regard those EEG changes as proven adverse effects, and its operational restrictions above 100 kHz remain substantially anchored to thresholds for RF-induced temperature rise or established stimulation effects.

The central disagreement is therefore not whether any nonthermal biological observation has ever occurred.

The disagreement is whether such observations are sufficiently replicated, mechanistically understood, and connected to adverse health outcomes to be incorporated into exposure-limit setting.

ICNIRP’s position is that current restrictions protect against substantiated adverse effects. Critics, including ICBE-EMF, argue that the standards exclude important chronic, nonthermal, modulation-dependent, reproductive, neurological, genotoxic, and carcinogenic evidence.

RF Safe’s position is that the standards ask too little of biology.

If an exposure changes a neural rhythm, calcium pattern, channel state, gene-expression program, reproductive endpoint, or tumor probability without measurable heating injury, that observation should not be dismissed merely because it has not yet been assigned to a single disease category.

Effect is not automatically injury. But effect is not irrelevance.

RF is one upstream modifier—not the only one

Low-fidelity biology is not an RF-exclusive model.

Living systems encounter upstream modifiers through many routes:

• inhaled through air;
• ingested through food and water;
• administered or injected through medical and pharmaceutical products;
• absorbed through heat, light, sound, mechanical energy, and electromagnetic fields.

Documented air pollutants, water contaminants, nutritional deficiencies, excessive ultra-processed-food exposure, endocrine-active chemicals, pharmaceuticals, infections, chronic stress, and other physical or chemical inputs can act on overlapping biological systems.

The common denominator is not the route of entry. It is the possibility of disturbing energy management, information processing, repair, development, or regulatory timing.

Scientific discipline is essential here. Something is not hazardous simply because it is injected, inhaled, artificial, or difficult to pronounce. Likewise, something is not safe merely because it is natural, familiar, or legally permitted.

Specific claims about an atmospheric intervention, food additive, pharmaceutical ingredient, vaccine component, contaminant, or RF waveform require specific evidence: identity, composition, dose, exposure route, timing, endpoint, susceptibility, and reproducibility.

The low-fidelity model does not validate every claim by analogy.

It supplies a common systems question:

Does this input measurably reduce biological fidelity in a defined receiver under defined conditions?

What makes ambient RF policy-distinctive

RF need not be declared the greatest environmental threat to be recognized as uniquely important in public policy.

Many exposures involve at least some identifiable decision: purchasing a product, eating a food, taking a medication, entering a workplace, or using a particular device.

Ambient wireless infrastructure is different. A person may reduce personal device use and still be exposed to signals originating from neighboring routers, utility infrastructure, nearby antennas, schools, workplaces, public systems, transportation networks, satellites, and other people’s equipment.

The magnitude is not identical everywhere, and no individual receives one constant “24/7 dose.” Nevertheless, the infrastructure can operate continuously, and meaningful individual opt-out may be impossible in many settings.

That is why consent matters.

Section 704 of the Telecommunications Act prohibits state and local governments from regulating the placement, construction, or modification of compliant personal wireless facilities on the basis of the environmental effects of RF emissions.

This converts compliance with FCC rules into a barrier against local health- and environmental-based decision-making—even though the D.C. Circuit subsequently found that the FCC had not adequately explained major aspects of those rules.

The NHF July 17 policy framework appropriately connects this issue to Section 704, wireless densification, local control, informed consent, and the availability of fiber and LiFi alternatives.

RF exposure is therefore distinctive not because every field is necessarily more dangerous than every chemical, pollutant, or pharmaceutical.

It is distinctive because it can be:

• ambient;
• persistent;
• difficult to avoid;
• imposed without an individual transaction;
• authorized through standards that do not account for every potentially relevant biological variable;
• protected from certain forms of local health-based regulation.

That combination creates an exceptional duty to prove safety with exceptional rigor.

Generation after generation

Today’s electromagnetic environment will not end with the current adult population.

Children are developing inside it. Pregnancies occur inside it. Schools and homes increasingly depend on it. Future adults may experience wireless exposure from conception through old age.

That fact does not prove inherited transgenerational injury. Evidence for germline transmission of RF-induced effects remains a research question.

But the persistence of exposure across successive generations creates several urgent study requirements:

• prenatal and placental exposure;
• embryonic and fetal bioelectric patterning;
• neural and immune developmental windows;
• puberty and reproductive maturation;
• germ-cell effects;
• interaction with genetic susceptibility;
• cumulative exposure history;
• whether parental exposure changes offspring phenotype or resilience.

A society should not treat children as the passive validation cohort for an exposure system whose biological assumptions remain contested.

The longer the deployment horizon, the stronger the obligation to study developmental timing and susceptible subgroups before harm becomes visible at population scale.

The evidence hierarchy: what is established

The RF Safe framework becomes strongest when it separates observations from interpretations.

Established empirical anchors

It is established that calcium signaling carries information through timing and oscillation frequency, not merely through average calcium concentration.

It is established that S4 voltage-sensor helices contain charged residues that detect the transmembrane electric field and control ion-channel gating.

It is established that certain RF exposures below established heating-injury thresholds can modify measurable human EEG activity. ICNIRP itself acknowledges effects in waking alpha and sleep-spindle frequency bands below 2 W/kg, while disputing that these constitute demonstrated harm.

It is established that the 2025 CACNA1C study reported a genotype-dependent human sleep-spindle response to a controlled 3.6 GHz exposure.

It is established that a defined EMF-inducible system can use CYB5B-dependent, rhythmic calcium dynamics to control gene expression.

It is established that specifically amplitude-modulated, very-low-level RF fields have been authorized by FDA for a limited therapeutic application on the basis of safety and probable benefit.

It is established that NTP reported clear evidence of malignant heart schwannomas and some evidence involving brain and adrenal tumors in male rats under its protocol.

It is established that the WHO-related 2025 animal review rated evidence as high certainty for increased gliomas and malignant heart schwannomas in male rats, while finding no or minimal evidence in most other organ systems.

It is also established that 2026 Japanese and Korean studies did not reproduce those tumor or genotoxicity findings at 4 W/kg under their CDMA protocol.

That is the empirical record. It is not a simple slogan in either direction.

What the evidence strongly implies

Several conclusions follow reasonably from those established observations.

First, SAR and temperature are necessary but incomplete descriptors of biological interaction. They quantify energy absorption and heating-related risk, but they do not fully encode modulation, pulse structure, timing, receiver state, or genotype.

Second, population averages can conceal susceptible subgroups. The CACNA1C result demonstrates that an effect absent in an aggregate population can appear after stratification by receiver architecture.

Third, waveform and timing deserve treatment as biological variables. Calcium signaling, EEG responses, therapeutic RF systems, and the CYB5B gene-switch findings all make this inference scientifically credible.

Fourth, a lack of one consistent Disease-X signal does not prove the absence of upstream biological activity. It may constrain the size or universality of a risk while leaving conditional and preclinical effects unresolved.

Fifth, current exposure limits were not designed around receiver-dependent low-fidelity biology. They are not genotype-specific, sleep-stage-specific, developmental-stage-specific, or modulation-specific.

These are not claims that every wireless exposure is harmful. They are reasons the existing safety question is incomplete.

What remains an RF Safe hypothesis

Several components of this manifesto are proposed research concepts rather than settled consensus.

It remains to be demonstrated whether ordinary telecommunications RF perturbs S4 channel gating directly under realistic exposure conditions.

It remains to be demonstrated whether CYB5B mediates responses to Wi-Fi, cellular, 5G, or other RF signals.

Density gating remains a hypothesis requiring quantitative validation.

Low-fidelity biology requires standardized biomarkers and prospective validation.

Entropic waste requires operational measurement and differentiation from ordinary biological variability.

The meta-disease state requires longitudinal evidence showing that measured fidelity loss predicts reduced resilience or multiple downstream pathologies.

ceLLM’s proposal that genomic and chromatin architecture functions as a weighted biological information network must be tested against alternative models.

These concepts should not be weakened by pretending they have already been fully proven.

Their strength lies in the fact that they integrate established observations into a coherent set of testable predictions.

The RF Safe research program

A serious research program should no longer expose cells, animals, or people to vaguely described “RF” and then search indiscriminately for any statistically significant endpoint.

It should begin with exact field characterization: carrier frequency, modulation envelope, pulse structure, duty cycle, peak and average intensity, polarization, harmonics, near-field geometry, exposure duration, repetition interval, and spatial distribution.

Thermal conditions must be continuously measured rather than assumed.

Experiments should be blinded, sham-controlled, preregistered, appropriately powered, independently replicated, and published with complete exposure files so another laboratory can reproduce the actual signal—not merely its nominal frequency.

Human studies should be stratified prospectively by genomic and physiological receiver variables, including CACNA1C and other ion-channel variants, epigenetic state, age, sex, developmental stage, circadian phase, sleep stage, medications, calcium-channel-blocker use, mitochondrial condition, and baseline autonomic or immune state.

Mechanistic studies should measure more than cell viability. Relevant endpoints include:

1. S4 and voltage-sensor gating kinetics;
2. calcium-waveform frequency, amplitude, phase, and compartment;
3. CYB5B dependence through knockout and rescue;
4. mitochondrial membrane potential and redox coupling;
5. transcriptional and epigenetic responses;
6. neural-network synchronization and sleep rhythms;
7. immune activation and resolution;
8. DNA damage, repair, apoptosis, and protein clearance;
9. developmental patterning and differentiation;
10. recovery after exposure.

Dose-response studies must test for nonlinear and windowed effects rather than assuming that every response increases smoothly with average power.

Longitudinal studies should ask whether repeated small perturbations recover fully, produce adaptation, create sensitization, or accumulate as repair debt.

Animal and organoid studies should explicitly examine prenatal and early-life windows. Human studies should stop relying predominantly on crude subscription or call-time proxies and incorporate personal dosimetry, device telemetry, network conditions, and waveform-resolved exposure where ethically and technically possible.

Finally, the field needs a validated Biological Fidelity Index: a composite measure of timing precision, state stability, recovery, and regulatory coherence. Such an index could allow researchers to detect an upstream disturbance before waiting decades for a tumor or chronic diagnosis.

What a modern exposure standard should measure

A next-generation safety framework should not abandon SAR. It should place SAR inside a larger dosimetric architecture.

Standards should consider:

• peak as well as average exposure;
• duty cycle and pulse repetition;
• low-frequency modulation embedded in RF carriers;
• cumulative duration and recovery intervals;
• simultaneous multi-source exposure;
• localized tissue geometry;
• developmental and reproductive windows;
• sleep and circadian exposure;
• genotype- and channel-defined susceptibility;
• long-term environmental exposure;
• endpoints below overt tissue injury.

Exposure standards should include explicit uncertainty factors for children and biologically susceptible subgroups rather than assuming that an averaged adult model represents everyone.

A compliance test should not merely determine whether a device can avoid exceeding a thermal threshold under a prescribed configuration. It should determine whether realistic use, aggregate exposure, and signal architecture have been adequately evaluated for biological effects.

A precautionary technology policy is not anti-technology

The choice is not connectivity or health.

The choice is whether connectivity will be engineered intelligently.

Fiber should form the physical backbone wherever feasible. Wired connections should remain available in homes, schools, hospitals, workplaces, and public facilities. LiFi and other optical systems can supply wireless-like mobility in appropriate settings without making RF the compulsory default for every data connection.

Wireless systems should be treated as a useful option whose deployment carries an obligation to minimize unnecessary exposure—not as an ambient entitlement that overrides local control and individual consent.

Practical exposure reduction does not require panic. It can include:

• distance;
• lower transmit power;
• wired backhaul and endpoints;
• disabling unused radios;
• reducing unnecessary nighttime transmission;
• thoughtful antenna siting;
• transparent exposure information;
• preservation of wired alternatives;
• special protection for sleeping, educational, medical, and developmental environments.

The goal is not zero electromagnetic interaction. Life itself is electromagnetic.

The goal is to avoid imposing unnecessary, poorly characterized, biologically active noise on systems whose signaling fidelity is essential to health.

The new scientific position

The old argument was:

Non-ionizing radiation cannot break chemical bonds directly, so exposures below significant heating are harmless.

That argument was always too narrow.

Biology is not affected only through direct bond ionization. It can be influenced through channel gating, calcium timing, membrane voltage, redox processes, neural synchronization, gene regulation, and other state-dependent mechanisms.

The opposite extreme is also scientifically unsound:

Every biological response proves disease.

A measurable response may be adaptive, neutral, therapeutic, transient, harmful, or harmful only in a susceptible receiver.

The scientifically durable position is:

Certain electromagnetic-field exposures can produce nonthermal biological effects. Whether a particular effect is adverse depends on the field pattern, biological receiver, timing, exposure history, compensatory capacity, and downstream consequence. Current exposure standards do not fully model that complexity.

That statement is forceful because it is accurate.

The manifesto

We do not need to claim that every wireless signal causes every disease.

We do not need to declare RF more dangerous than every pollutant, chemical, nutritional insult, pharmaceutical exposure, or environmental intervention.

We do not need to turn uncertainty into certainty or correlation into causation.

We need to stop pretending that the absence of immediate heat injury proves biological silence.

We need to stop treating the average human as the only human.

We need to stop averaging away the timing structures that living systems may detect.

We need to stop demanding a single downstream diagnosis before acknowledging an upstream disturbance.

We need standards that recognize that calcium is a time code, that voltage sensors are molecular field detectors, that genotype changes receiver behavior, that signal patterns can control gene expression, and that experimental animals have produced both serious positive findings and important negative replications.

We need research capable of explaining why.

We need public policy that distinguishes compliance from safety and deployment authority from informed consent.

We need a model that understands biological health not merely as the absence of a named disease, but as the preservation of signaling precision, adaptive capacity, repair, and coherence.

That model is high-fidelity biology.

Its degradation is low-fidelity biology.

The accumulated unresolved burden is entropic waste.

The conditional architecture of susceptibility is receiver-dependent biology.

The state in which multiple regulatory systems lose resilience before a conventional diagnosis appears is the meta-disease state.

The hypothesis that target density and biological organization determine whether a signal crosses a response threshold is density gating.

And the ceLLM proposition is that the cell is not merely reading a static blueprint. It is continuously resolving biological outputs through a weighted, state-dependent information architecture.

These terms are not substitutes for evidence.

They are a way to organize evidence that the old vocabulary has failed to connect.

The truth we can now place our finger on

The verified truth is not that every RF exposure causes Disease X.

The verified truth is that living systems can detect and respond to selected electromagnetic-field patterns without thermal injury.

The verified truth is that calcium timing carries biological information.

The verified truth is that cells contain molecular voltage sensors.

The verified truth is that a human genetic variant can alter the measurable brain response to the same controlled RF exposure.

The verified truth is that a defined electromagnetic field can activate a CYB5B-dependent calcium rhythm and control gene expression in an engineered biological system.

The verified truth is that weak, structured RF fields can be used therapeutically.

The verified truth is that major animal studies have produced serious cancer signals, that a WHO-related systematic review judged two of those animal outcomes with high certainty, and that subsequent studies have also produced important negative results.

The verified truth is that the FCC failed to justify its refusal to revisit central questions concerning children, long-term exposure, modulation, technological change, and environmental effects.

The verified truth is that an exposure framework built primarily around preventing excessive heating cannot be assumed to protect every dimension of biological signal integrity.

What remains is to determine the boundary between detectable effect, adaptive response, cumulative fidelity loss, and disease.

That boundary will not be found by returning to the same thermal-only operational assumptions.

It will be found by measuring the field more completely, measuring the receiver more completely, and measuring biology upstream—where timing first changes.

The future of RF science is not simply energy accounting.

It is signal-integrity science.

The receiver is part of the dose.

Timing is part of the mechanism.

Fidelity is part of health.