How an ICNIRP-linked interpretation became a wireless-safety verdict—and why NTP, Ramazzini, nonlinear responses, S4 dynamics, CACNA1C and CYB5B point toward the question thermal standards still have not tested: low-fidelity biology

A verdict arrived before the institution responsible for issuing the verdict had written it.

On July 17, 2026, Medical Xpress announced: “No evidence mobile phones cause brain cancer—new study.” The following day, the Times of India went further, promising that a “new study” had finally delivered a clear answer.

It had not.

The Times of India incorrectly described the source as a new Australian study, said researchers had reviewed “everything out there,” misidentified the journal in which it appeared and told readers that the answer was now “loud and clear.” Medical Xpress was more careful: it acknowledged that the World Health Organization had not completed its analysis of the commissioned reviews. Yet it still presented the authors’ interpretation as a preview of WHO’s expected conclusion and assured readers that current exposure limits were “likely safe.”

The actual source was a two-author narrative overview by Mark Elwood and Ken Karipidis in the New Zealand Medical Journal. It was not a new experiment, a new systematic review, a new meta-analysis or WHO’s final health-risk assessment. The authors’ own paper states that a separate WHO Task Group must still review the evidence, formulate the integrated risk assessment, consider interventions and identify research gaps. WHO describes exactly the same unfinished process.

That distinction is not editorial trivia. It is the difference between reporting evidence and manufacturing finality.

The Times of India and Medical Xpress should therefore not be trusted as accounts of a final WHO judgment. This does not mean that their writers fabricated the underlying research. It means that they converted a qualified, institutionally situated interpretation into a public safety verdict that does not yet exist.

That process has a name:

Conclusion laundering.

A review finds reassurance for several specified outcomes. A narrative overview selectively organizes the review portfolio around that reassurance. Media outlets call the overview a “new study.” The uncertainty disappears. The institutional relationships disappear. The contradictory animal evidence becomes a footnote. And an unfinished WHO assessment re-emerges as a headline saying the controversy has been settled.

It has not.

---

This was not an independent verdict from outside the standards system

The institutional issue cannot remain buried in a competing-interests paragraph.

Ken Karipidis was a member of the WHO group that conducted the human observational cancer reviews. He is also ICNIRP’s vice-chair and chairs ICNIRP’s Standing Committee on Communication. Dan Baaken, another author of the WHO human-cancer reviews, is ICNIRP’s scientific secretary, a member of its board and vice-chair of the same communications committee. Young Hwan Ahn, an investigator in the Japan–Korea animal project, joined the ICNIRP commission in 2024; ICNIRP’s biography specifically highlights his participation in what it calls the Japan–Korea “NTP validation” study.

ICNIRP is not a marginal scientific society. Its exposure guidelines form the basis of national standards in numerous countries, and WHO has formally recognized its role in international non-ionizing-radiation protection.

The Korean investigators explicitly said that they selected a whole-body SAR of 4 W/kg because it was the animal reference point used to establish human RF-exposure limits. ARPANSA later cited the absence of a statistically significant tumor increase at that same guideline-centered exposure as support for ICNIRP’s existing safety assessment.

That produces a closed validation loop:

The standards network establishes the thermal reference point, helps conduct or frame evidence testing at that point, evaluates the resulting literature, communicates the interpretation and then cites the interpretation as validation of the standards.

This is not proof of falsified data, secret coordination or conscious dishonesty. Those allegations would require evidence that is not presently available.

It is, however, a serious structural and intellectual conflict of interest.

Financial payments are not the only source of motivated reasoning. Scientists and institutions that have written, defended and communicated a regulatory framework acquire professional, reputational and policy commitments to that framework. Reversing course would require acknowledging that the protection system they helped construct may have omitted biologically relevant variables.

The conflict is also not symmetrical.

RF-safety critics may have advocacy commitments, and their methods should receive the same scrutiny as anyone else’s. But they do not write the internationally adopted exposure guidelines. They do not operate the official communications committee for those guidelines. They are not evaluating a policy architecture that they themselves helped institutionalize.

The point is not that ICNIRP participation automatically invalidates a paper.

The point is that a standards-setting network must not be mistaken for an independent external judge of evidence challenging its own standards.

---

Even the WHO review portfolio did not return “zero”

The media narrative also fails on its own evidentiary terms.

Elwood and Karipidis’s overview reports relatively reassuring human epidemiology for several heavily investigated brain and head tumors. That is important evidence and should be acknowledged.

But the same overview also reports:

• High-certainty evidence of increased glioma and malignant cardiac schwannoma in male rats.
• Moderate-certainty evidence involving several additional animal tumors.
• Experimental reproductive findings including reduced fetal weight, pregnancy rate and sperm count.
• Numerous human and mechanistic outcomes supported by low- or very-low-certainty evidence, meaning that those questions remain poorly resolved rather than affirmatively cleared.

The WHO-commissioned animal-cancer review included 52 studies. It judged the evidence to be of high certainty for increased glioma and malignant heart schwannoma in male rats. The review has been criticized over its synthesis, risk-of-bias judgments and evidence grading. Its authors subsequently published a direct methodological defense, and further correspondence continued the debate. That is an active scientific dispute—not a retraction and not a successful erasure of the animal findings.

Elwood and Karipidis emphasize the criticisms of the animal review, describing it as heavily criticized, but their references do not provide readers with an equivalent account of the review authors’ published defense.

“Dismissed” may be too strong a word.

Down-weighted and selectively presented is not.

The unresolved task for WHO is not to count how many review summaries contain reassuring language. It is to explain how moderately reassuring human epidemiology should be integrated with high-certainty positive animal evidence, experimental reproductive findings, uncertain mechanistic evidence and major deficiencies in exposure characterization.

That integrated judgment remains pending.

---

Brain-cancer epidemiology matters—but it is not the whole RF question

The strongest RF Safe position does not require pretending that every human cancer study is positive.

Prospective cohorts, incidence analyses and systematic reviews have made a large population-wide increase in several heavily studied brain and head tumors less plausible over the exposure histories and latency periods examined. That is real reassurance.

But reassurance against a large average effect in selected diagnoses is not proof that:

• Smaller effects do not exist.
• Long-latency effects do not exist.
• Highly exposed or biologically susceptible subgroups do not exist.
• Exposure misclassification has not diluted associations.
• Modern waveforms are biologically interchangeable with older ones.
• Non-cancer effects have been adequately tested.
• Upstream biological regulation remains unaffected.

The Danish mobile-phone cohort illustrates the problem. It classified exposure by possession of a qualifying private subscription, excluded more than 200,000 corporate subscriptions from the nominally exposed cohort and allowed later mobile-phone users to accumulate in the comparison population. It lacked individual information on call duration, laterality, device technology, output power and cordless-phone use. Those limitations severely reduced the exposure contrast on which its reassuring result depended.

The scientifically responsible synthesis is therefore not “human evidence proves harm,” but neither is it “human evidence proves biological neutrality.”

It is this:

Human epidemiology increasingly argues against a large average increase in several specified brain tumors. It remains much less capable of resolving small effects, long latency, waveform specificity, childhood exposure, exposure misclassification and susceptible minorities.

Most importantly, brain cancer is a downstream diagnosis. It is not a direct measurement of ion-channel timing, calcium-wave precision, mitochondrial reserve, redox recovery, immune surveillance or circadian organization.

The red herring is not studying brain cancer.

The red herring is treating brain cancer as a referendum on every possible biological consequence of RF exposure.

---

What the 2026 Japan–Korea studies actually tested

The protocol comparison is decisive.

NTP used multiple exposure groups and two modulations. Ramazzini examined a vastly lower far-field exposure regime. Japan and Korea tested one modulation, one sex and one SAR positioned between NTP’s 3 and 6 W/kg groups.

The Japanese authors expressly stated that their study was not intended as a direct replication. The Korean authors said it was not designed as a complete replication. ARPANSA itself calls the experiments “partial replication studies.”

That does not make the studies worthless. Their GLP procedures, exposure characterization, pathology review and blinding are significant strengths.

But GLP answers a limited question:

Was the selected experiment executed faithfully?

It does not answer:

Was the selected experiment capable of testing the competing biological hypothesis?

A GLP-perfect single point is still a single point.

One exposure level cannot define a dose–response curve. It cannot identify a non-monotonic response. It cannot distinguish a biologically inactive point from one at which stress and compensation coexist. One modulation cannot test modulation specificity. A male-only design cannot test sex-dependent responses. A high, guideline-centered SAR cannot test the Ramazzini low-SAR regime.

And terminal tumor counts cannot reconstruct transient oxidative stress, altered calcium timing, adaptive signaling, mitochondrial reserve or recovery kinetics occurring months before a tumor is diagnosed.

---

The null was real. The generalization was not.

The correct conclusion from the new studies is:

Under one intermittent 900 MHz CDMA protocol at a nominal whole-body SAR of 4 W/kg, two male-rat experiments did not detect a statistically significant increase in tumor incidence.

That is a legitimate result.

It is not equivalent to:

RF exposure does not cause cancer in rats.

Yet the second statement became ARPANSA’s headline and then entered the media pipeline as generalized reassurance. ARPANSA simultaneously acknowledged that the studies were partial replications, used only one exposure level, were underpowered for small increases in rare tumors such as glioma and did not reproduce NTP’s dose structure.

These were also not literally tumor-free experiments.

When the two studies were combined, the German Federal Office for Radiation Protection, BfS, counted:

• Three heart schwannomas in exposed animals versus one in sham animals.
• Two gliomas in exposed animals versus one in sham animals.

The differences were not statistically significant. But BfS noted that the estimates from both experiments pointed in the same direction, with very wide confidence intervals. It concluded that the findings were not necessarily inconsistent with NTP and that effects on specific heart and brain tumors could not be ruled out. BfS also emphasized that the smaller groups—70 animals rather than NTP’s 90—and the single exposure level limited the experiments’ ability to detect rare tumors and evaluate the exposure trend.

That is a very different scientific message from “RF does not cause cancer in rats.”

The null was real.

The generalization was not.

---

Why the choice of 4 W/kg was not scientifically neutral

The Korean paper explains that 4 W/kg was selected because it is the animal reference point from which human exposure limits were derived. ICNIRP treats a whole-body average SAR of 4 W/kg, averaged over 30 minutes, as the operational adverse-effect threshold associated with an approximately 1°C increase in body-core temperature. Reduction factors then produce whole-body restrictions of 0.4 W/kg for occupational exposure and 0.08 W/kg for the general public.

The Japan–Korea project therefore did not independently map the exposure landscape to discover where responses occurred.

It selected a single coordinate because that coordinate already held a privileged place in the existing guideline framework.

Then the result at that coordinate was used to support the framework.

That is regulatory circularity even without any allegation of misconduct.

It also matters because 4 W/kg is not demonstrably biologically neutral.

In a separate rat experiment using 2.14 GHz W-CDMA, Ohtani and colleagues reported core-temperature increases of roughly 1–1.5°C during prolonged 4 W/kg exposure, together with changes in several heat-shock-protein and heat-shock-factor genes. Comparable changes were not observed at 0.4 W/kg. The frequency and exposure schedule differed from the Japan–Korea studies, so this does not prove that their animals activated the same pathways. It does demonstrate that 4 W/kg can occupy a physiologically active region rather than an inert testing point.

The chronic Japan–Korea experiments themselves produced evidence of altered whole-animal physiology. Exposed animals generally consumed less food, weighed less, developed less chronic progressive nephropathy and, in the Japanese experiment, survived longer than sham animals. BfS proposed that rats housed below thermoneutrality may have required less dietary energy for body heating because RF energy deposition contributed to their thermal budget.

That is not “no biological effect.”

It is evidence that the exposure affected whole-animal energetics strongly enough to change food consumption, weight, renal pathology and survival—variables that can also affect tumor development and detection.

The scientifically supportable boundary-layer hypothesis is therefore:

Four watts per kilogram is a guideline-centered, physiologically active region in which thermal loading, metabolic adaptation and stress-response pathways may affect the observed outcome.

The stronger claim—that the investigators deliberately selected 4 W/kg because they knew it would guarantee a null—is not established.

Nor has it been shown that heat-shock or antioxidant responses erased an underlying carcinogenic effect.

The chronic studies did not report the longitudinal measurements needed to answer that question: HSP70, HSF1, Nrf2, glutathione balance, antioxidant enzymes, mitochondrial reserve, calcium dynamics and time-resolved ROS.

The decisive formulation is:

The protocol did not guarantee a null. It guaranteed that the boundary-layer hypothesis could not be tested.

---

NTP was not erased

NTP exposed rats to both GSM- and CDMA-modulated 900 MHz signals at 1.5, 3 and 6 W/kg.

Its final peer-reviewed conclusions found clear evidence of carcinogenic activity for malignant heart schwannomas in exposed male rats under both modulations. Malignant gliomas were considered related to exposure at NTP’s “some evidence” level. “Clear evidence” and “some evidence” are formal NTP categories—not casual adjectives and not direct estimates of human risk.

The raw GSM male-rat glioma incidences were:

0 → 3 → 3 → 2 across sham, 1.5, 3 and 6 W/kg.

Malignant heart schwannomas were:

0 → 2 → 1 → 5.

For CDMA, heart schwannomas were:

0 → 2 → 3 → 6.

These data do not prove a specific adaptive mechanism, and NTP reported statistically significant exposure trends after its survival-adjusted analyses. But the raw GSM tumor counts illustrate why biological monotonicity cannot simply be assumed.

Japan and Korea tested only CDMA at 4 W/kg.

That does not invalidate their study. It means their experiment could not answer whether GSM produced a different dose-response shape, whether a mid-dose GSM window existed, or whether waveform structure altered the response.

It is also inaccurate to portray NTP’s CDMA arm as biologically empty. NTP’s strongest heart finding occurred under both GSM and CDMA.

The precise criticism is:

The new studies tested one of NTP’s two modulations at one intermediate SAR. They did not reproduce NTP’s dose structure and could not test modulation-dependent nonlinearity.

---

Ramazzini occupies a parameter space the new studies never entered

The Ramazzini Institute exposed 2,448 male and female rats to a 1.8 GHz GSM far-field signal for 19 hours per day, beginning prenatally and continuing until natural death.

Its approximate whole-body SARs were 0.001, 0.03 and 0.1 W/kg—orders of magnitude below NTP and far below the Japan–Korea 4 W/kg protocol. At the highest Ramazzini exposure, the investigators reported a statistically significant increase in heart schwannomas in male rats, with additional non-significant glial and Schwann-cell findings.

Ramazzini does not establish that ordinary environmental exposure causes human cancer. Its findings have their own methodological and interpretive limitations.

But a 4 W/kg CDMA-only experiment cannot falsify a result observed under a far-field GSM protocol at approximately 0.1 W/kg or below.

The exposure spaces barely overlap.

A study conducted at 4 W/kg cannot settle what occurs at 0.1 W/kg any more than a drug trial using one high dose can define the complete response at doses forty, four hundred or four thousand times lower—particularly where the hypothesis specifically concerns non-monotonicity.

The WHO-commissioned animal review’s high-certainty judgments were driven substantially by the convergence of the NTP and Ramazzini findings involving glial tumors and cardiac schwannomas under very different exposure regimes. Whether those findings translate quantitatively to humans remains unresolved. Their existence is not.

---

The research supports nonlinearity more strongly than it supports one universal mechanism

A biological system is not obliged to respond in a straight line.

Stress, injury, compensation, adaptation, repair and failure can overlap. A response may appear at one intensity and diminish at another. It may peak during exposure and disappear by the time tissue is collected. A higher average dose may activate pathways that a lower dose does not. A long exposure may produce a different state from repeated short exposures even when the integrated absorbed energy is similar.

That does not make every irregular result evidence of a hidden RF effect. Random variation, poor dosimetry, small samples, multiple comparisons and publication bias can also produce irregular patterns.

It means only that one point cannot establish linearity, monotonicity or biological inactivity.

The Durdik umbilical-cord-blood experiment provides a useful example of the timing problem. At a reported SAR of 40 mW/kg, the researchers detected an average ROS increase of about 30% after one hour of UMTS exposure, with substantial interindividual variation. The increase was not detected after three hours, and the investigators found no persistent DNA damage, preleukemic fusion genes or apoptosis under their conditions. They also cautioned that changing oxygen concentration during incubation could have contributed to the difference.

Durdik therefore supports a narrow but important conclusion:

RF-associated cellular responses can be time-dependent, and a study can miss a transient response by measuring only after it has changed or resolved.

It does not prove that a repair switch activated and erased an injury.

That missing causal sequence would require simultaneous, time-resolved measurements of ROS generation, antioxidant activity, calcium dynamics, HSP and Nrf2 signaling, mitochondrial respiration, DNA damage and recovery.

The 2025 Jamaludin Wi-Fi study is also consistent with a biphasic interpretation: oxidative and reproductive measures were most abnormal in the four-hour group and appeared to improve in the eight- and twenty-four-hour groups. But it involved only six rats per group, used a commercial router 20 centimeters from the cages, reported no usable whole-body or tissue SAR and printed Pearson correlation coefficients of −9.69 and −9.55—mathematically impossible values because a correlation coefficient must lie between −1 and +1.

Jamaludin is hypothesis-generating evidence.

It should not be a pillar of the boundary-layer argument.

The stronger foundation is the convergence of:

• NTP’s multi-dose and multi-modulation findings.
• Ramazzini’s low-SAR far-field signal.
• Ohtani’s evidence that 4 W/kg can engage thermal-stress physiology.
• The altered food intake, weight, nephropathy and survival seen in chronic studies.
• Durdik’s demonstration that endpoint timing can determine whether a cellular response is detected.

Together, these findings do not prove one universal non-thermal mechanism.

They do make it scientifically indefensible to assume that every biologically relevant response must be linear, monotonic, persistent and fully represented by terminal tumor counts.

---

ICNIRP’s averaging framework may discard the variable carrying the biological information

ICNIRP’s RF limits are fundamentally organized around absorbed energy, temperature and avoidance of established thermal effects.

ICNIRP acknowledges the question of whether continuous and pulsed exposures might produce different biological effects. But its guidelines state that no theoretical distinction is made between those exposure forms within its health-protection framework. Whole-body SAR is averaged over 30 minutes, while local SAR is generally averaged over six minutes.

That approach is logical if average absorbed energy and temperature are the only biologically material variables.

It may be systematically incomplete if response also depends on:

• Pulse interval.
• Duty cycle.
• Rise and fall time.
• Amplitude modulation.
• Frequency transitions.
• Peak-to-average ratio.
• Polarization and field geometry.
• Temporal coherence.
• Burst repetition.
• The phase relationship between exposure and endogenous oscillations.

Averaging is not neutral.

Averaging intentionally removes temporal structure.

That is appropriate when temporal structure is irrelevant. But if timing is causal, averaging can remove the variable that carries the biological information.

Consider two signals with the same average SAR. One is continuous. The other delivers energy in sharply organized bursts. A thermal-average standard may regard them as equivalent after the selected averaging period.

A timing-sensitive biological receiver may not.

The unresolved question is not whether the average-power calculation is mathematically correct.

It is whether average power is biologically complete.

---

A biological organism is not a calorimeter

Living systems are electrochemical timing networks.

The heart depends on precisely ordered depolarization and repolarization. Neurons encode information through spike timing and synchrony. Calcium operates through oscillations, waves and pulses—not merely through an average intracellular concentration. Mitochondria couple calcium timing to ATP production and redox state. Hormones are secreted in circadian, ultradian and pulsatile patterns. Immune signaling depends on sequence, duration and timing. Development depends on events occurring within specific windows.

In such systems, timing is information.

A small increase in timing noise need not destroy a pathway. It may reduce precision. A calcium wave may still occur but with greater phase jitter. A membrane channel may still function but with altered opening probability or recovery. A mitochondrion may still produce ATP but with diminished reserve. A stress response may still restore baseline, but more slowly or incompletely.

That is the conceptual space in which S4 dynamics, CACNA1C and CYB5B become relevant.

Not because any one of them has already proved a universal mechanism of wireless disease.

Because each points toward a body capable of receiving electromagnetic input through timing-sensitive molecular machinery.

---

S4 dynamics: a plausible transduction hypothesis, not settled physics

Voltage-gated ion channels contain charged voltage-sensing structures. In many channels, the positively charged S4 helices move in response to changes in membrane potential and help control channel opening and closing.

The ion-forced-oscillation/voltage-gated-ion-channel model advanced by Panagopoulos and colleagues proposes that structured electromagnetic fields can move mobile ions near these voltage sensors, exert additional forces on the charged S4 segments and disturb channel gating. Ionic dysregulation could then alter calcium signaling, mitochondrial activity, NADPH oxidases and reactive-oxygen production.

The model is attractive because it identifies a candidate route from weak, temporally structured fields to amplified cellular consequences.

It is also still a proposed mechanistic synthesis.

The 2025 Panagopoulos paper is a review and theoretical argument, not a direct demonstration that ordinary Wi-Fi, LTE or 5G exposure forces S4 gating and causes disease. Some of its broader claims—for example, universal descriptions of anthropogenic fields or the proposition that embedded low-frequency components explain essentially all non-thermal RF effects—go beyond what controlled experiments have established.

RF Safe’s strongest defensible position is therefore not:

S4 has been proved to explain every RF-associated disease.

It is:

S4/VGIC timing is a biologically plausible transduction pathway that thermal-average standards do not directly test, and the necessary waveform-controlled, electrophysiological experiments have not been performed at sufficient scale to exclude it.

A decisive test would hold carrier frequency, average absorbed power, peak field, exposure duration and temperature constant while changing only the waveform. It would measure channel currents directly by patch clamp, quantify activation and recovery kinetics, alter the S4 sensor through targeted mutations and determine whether blockers or genetic interventions abolish the RF-associated response.

Until that work is done, S4 should be neither declared proven nor waved away as impossible.

---

CACNA1C: evidence that the receiver can change the response

A small 2025 randomized, double-blind, sham-controlled study examined 34 participants genotyped for a variant in CACNA1C, the gene encoding the α1C subunit of an L-type voltage-gated calcium channel.

Participants received standardized 700 MHz and 3.6 GHz exposures before sleep. The researchers reported that 3.6 GHz exposure altered sleep-spindle center frequency in one CACNA1C genotype group but not in the matched comparison group. The experiment did not demonstrate disease or clinical harm, and its small sample requires independent replication.

Its importance lies elsewhere.

Identical experimental exposure did not produce an identical measured response across genotypes.

That is a concrete example of a central low-fidelity principle:

The receiver is part of the dose–response.

Conventional dosimetry must still quantify frequency, field strength, power density and SAR. But absorbed dose alone may not predict response if channel expression, genotype, developmental stage, membrane state, metabolic health, circadian phase or baseline inflammation modify biological sensitivity.

Population averaging can then conceal a real response concentrated within a minority.

A null average does not establish that every receiver was unresponsive.

---

CYB5B: proof that electromagnetic input can be translated into calcium timing

A 2026 Cell paper developed an engineered electromagnetic-field-responsive gene switch for remote control of gene expression.

Using a CRISPR screen, the investigators identified CYB5B as an essential mediator of the engineered response. Most significantly, activation depended on rhythmic calcium oscillations, not merely a generic increase in cellular calcium.

This was a deliberately constructed biomedical system.

It is not evidence that ordinary wireless exposure uses CYB5B to cause cancer, infertility or neurological disease.

But it is an important proof of principle:

Under defined conditions, electromagnetic input can be converted through a specific molecular mediator into a temporally organized calcium signal that controls gene expression.

That observation matters because it demolishes the crude assumption that the only biologically meaningful endpoint of electromagnetic exposure must be bulk heating.

It does not prove that the engineered pathway operates under environmental RF conditions.

It proves that biological electromagnetic transduction and timing-coded calcium responses are physically and molecularly possible.

The next questions are experimental:

• Does CYB5B respond under ordinary RF exposure conditions?
• Does its response depend on carrier frequency, modulation or field geometry?
• Does knocking down CYB5B alter any reproducible RF-associated endpoint?
• Does it interact with voltage-gated calcium-channel activity?
• Are its calcium oscillations adaptive, disruptive or biologically neutral outside the engineered gene-switch system?

Those questions belong in the research program—not outside the permissible boundaries of the safety model.

---

Low-fidelity biology: the upstream hypothesis

RF radiation does not need to map one-to-one onto one disease in every exposed organism.

That expectation may itself be the category error.

RF Safe’s proposed alternative is low-fidelity biology: a systems-level state in which regulatory networks continue to function but do so with reduced timing precision, signal discrimination, synchronization or recovery capacity.

The proposed sequence is:

Structured electromagnetic exposure
× receiver susceptibility
× developmental and physiological state
× co-exposures
→ altered channel or signaling timing
→ ionic and mitochondrial disturbance
→ redox and stress response
→ incomplete or altered recovery
→ reduced biological fidelity
→ changed susceptibility to downstream disease

Within this framework:

Biological dissonance is the perturbing process.

Low-fidelity biology is the state of reduced regulatory precision.

A meta-disease state is the resulting reduction in resilience across multiple systems.

A specific diagnosis is downstream geography.

Low-fidelity biology is not currently a recognized medical diagnosis. A meta-disease state is not an established clinical entity. They should not be advertised as such.

They are proposed, testable systems-biology concepts.

The central model can be expressed as:

Biological response = exposure × waveform × timing × genotype × developmental stage × tissue state × co-exposures × baseline resilience

Under this model, two people receiving similar external exposure need not develop the same outcome.

One may show no measurable change.

Another may exhibit a transient sleep or autonomic effect.

A third, with a particular channel genotype, reduced mitochondrial reserve, chronic inflammation or concurrent environmental stress, may recover less efficiently.

The theory does not predict that RF exposure inevitably causes a disease.

It predicts that responses should cluster around identifiable physiological and genetic characteristics, that waveform and timing should matter, and that altered recovery should sometimes be more informative than the largest instantaneous change.

---

What a low-fidelity state would look like experimentally

A meaningful theory must generate measurements that can fail.

Low-fidelity biology predicts outcomes such as:

• Increased phase jitter in calcium oscillations.
• Greater cell-to-cell variability under identical stimulation.
• Altered channel activation, inactivation or recovery kinetics.
• Reduced signaling dynamic range.
• Delayed return to baseline after exposure.
• Altered coupling between calcium signaling and mitochondrial metabolism.
• Reduced mitochondrial reserve capacity.
• Persistent redox imbalance after the field is removed.
• Greater damage from a standardized second stressor.
• Genotype-dependent response distributions.
• Loss of circadian or ultradian synchronization.
• Impaired coordination of DNA repair, apoptosis or immune clearance.

The theory would be weakened if rigorous studies repeatedly found:

• No waveform dependence after temperature and absorbed power were controlled.
• No reproducible calcium-timing or channel-current changes.
• No altered recovery.
• No susceptibility predictors.
• No interaction with a second stressor.
• No abolition of the response after disrupting the proposed transduction pathway.

That possibility of failure is essential.

A theory that treats every positive result as confirmation and every negative result as proof that the experiment was inadequate is not falsifiable.

RF Safe does not need an unfalsifiable theory.

It needs a research program capable of distinguishing biological timing effects from heating, artifacts and random noise.

---

What is demonstrated—and what is not

Demonstrated

The 2026 Japan–Korea studies were partial, one-dose, CDMA-only experiments in male rats.

They did not detect statistically significant tumor increases at 4 W/kg.

Their pooled tumor counts were not uniformly zero, and BfS concluded that specific heart and brain effects could not be excluded.

NTP found clear evidence of malignant heart schwannomas in exposed male rats and some evidence involving malignant gliomas.

Ramazzini reported a significant male-rat heart-schwannoma finding under a much lower far-field GSM exposure.

The WHO-commissioned animal review judged the evidence for male-rat glioma and heart schwannoma to be of high certainty.

Four watts per kilogram can be physiologically active under some rat exposure protocols.

Transient and genotype-dependent responses have been observed in the Durdik and CACNA1C experiments.

An engineered CYB5B-dependent system has converted electromagnetic input into timing-coded calcium oscillations and gene expression.

Plausible but unresolved

RF biological responses may be non-monotonic.

Stress adaptation or metabolic compensation may alter chronic outcomes around 4 W/kg.

Pulse structure or modulation may matter independently of average SAR.

S4 voltage sensors may provide one route of electromagnetic transduction.

Susceptible genotypes or physiological states may respond differently from population averages.

Repeated timing disturbance may reduce biological resilience even when no single terminal endpoint crosses a statistical threshold.

Not demonstrated

The Japan–Korea investigators deliberately selected a protocol to manufacture a null.

A heat-shock or antioxidant response guaranteed that tumors would be suppressed.

GSM is the only biologically active wireless modulation.

S4 disruption is the universal initiating event for every reported RF effect.

CYB5B mediates disease from ordinary consumer wireless exposure.

Low-fidelity biology is already a confirmed clinical condition.

The distinction between these categories does not weaken the RF Safe position.

It protects it from easy dismissal.

---

The experiment that would genuinely test the competing hypothesis

The next animal and mechanistic program should not be organized around one privileged point in an existing guideline.

It should be factorial, waveform-resolved and explicitly designed to test timing.

1. Map the dose space

Include the Ramazzini environmental range and multiple higher exposures—for example, approximately 0.001, 0.03, 0.1, 0.5, 1.5, 3, 4 and 6 W/kg where technically and ethically appropriate.

A curve cannot be inferred from one point.

2. Separate waveform from absorbed energy

Use matched continuous-wave, GSM, CDMA and contemporary LTE/5G-derived signals while independently controlling:

• Average SAR.
• Peak SAR.
• Duty cycle.
• Burst timing.
• Peak-to-average ratio.
• Field geometry.
• Temperature.

The waveform data and exposure code should be published so independent laboratories can reproduce the exact signal.

3. Separate thermal loading from RF-specific effects

Include continuous core-temperature monitoring, thermoneutral and conventional housing, pair-fed controls and sham groups matched for handling and environmental conditions.

This would test whether changes in food intake, weight, renal disease or survival are mediated by RF energy deposition, caloric restriction or another pathway.

4. Resolve the biology in time

Measure during exposure, immediately after exposure and across recovery:

• Voltage-gated channel currents.
• Calcium-wave frequency, phase and amplitude.
• ROS and lipid peroxidation.
• Glutathione balance.
• Nrf2 signaling.
• HSP70 and HSF1.
• Antioxidant enzymes.
• Mitochondrial respiration and reserve.
• 8-OHdG and γH2AX.
• Apoptosis and senescence.
• Inflammatory and immune-surveillance markers.

A terminal measurement cannot explain a transient process.

5. Test the proposed transducers

Use channel blockers, targeted S4 modifications, CACNA1C genotypes and CYB5B perturbation where biologically appropriate.

The objective is not merely to observe correlation but to test causation:

Waveform
→ channel or transducer response
→ altered calcium timing
→ mitochondrial or redox response
→ altered recovery
→ changed long-term susceptibility

6. Design for heterogeneity

Include both sexes, prenatal initiation, lifetime follow-up, adequate power for rare tumors and prospective stratification by genotype and baseline metabolic state.

The receiver must be characterized rather than averaged out.

7. Make institutional independence part of the protocol

Preregister the hypotheses and analyses.

Use multiple independent exposure laboratories and pathology teams.

Publish null and positive findings together.

Include investigators with materially different interpretations of the existing evidence.

No standards-setting network should be the sole designer, evaluator and communicator of the experiment used to validate its own framework.

---

The real verdict

Times of India and Medical Xpress did not report WHO’s final judgment.

They amplified an interpretation produced by authors embedded in the evidence-review and exposure-standard ecosystem, then stripped away the unresolved animal findings, study-design limitations and institutional context needed to evaluate that interpretation.

The 2026 Japan–Korea studies are valuable.

They establish a boundary condition:

At 4 W/kg of intermittent 900 MHz CDMA exposure, the individual male-rat studies did not detect a statistically significant tumor increase.

They do not erase NTP.

They do not erase Ramazzini.

They do not test the low-SAR region.

They do not test GSM.

They do not test a non-monotonic curve.

They do not establish that 4 W/kg is physiologically neutral.

They do not test calcium timing, S4 dynamics, CACNA1C susceptibility, CYB5B transduction, mitochondrial recovery or low-fidelity biology.

And they do not validate the sufficiency of a framework that averages away the very timing variables at issue.

The most important question is no longer merely:

Does RF exposure map directly onto Brain Cancer X?

The deeper question is:

Can structured electromagnetic exposure add timing noise to biological control systems, lowering fidelity and resilience in susceptible receivers long before one specific disease becomes visible?

The research does not yet provide a final answer.

But neither do ICNIRP’s thermal averages.

That is the hypothesis the standards system has not tested.

That is the parameter space the reassuring headlines ignore.

And that is why a single guideline-anchored null cannot close the RF debate.

The burden on RF Safe is not to claim that every uncertain finding proves harm.

The burden on the standards system is to test the biological variables it currently averages away.