Summary: Most conversations about EMF safety ask whether a given signal is strong enough to cause harm. A viral X thread in March 2026 reframed that question in a way that got 4 million views: what if the government's own cancer study showed RF radiation was actually good for you? The thread was wrong, but understanding why it was wrong requires going much deeper than the data it cited. This piece covers what the 2018 NTP study actually found, why the animals in the control group died early, why that had nothing to do with RF being beneficial, and why the cancer signal, which peaked at the lowest dose tested, is the finding that should concern us. It also covers why standard statistical tools consistently fail to detect biological effects in systems governed by nonlinear dynamics, the researchers who proved that failure was systematic, and what a 2026 reanalysis by the scientist who designed the original NTP study found when he applied conventional risk modeling to the data. The regulatory infrastructure built to study this question has been shut down. The exposure limits in place were written when the Soviet Union still existed. The question of what the electromagnetic environment is doing to human biology has not been answered. It has been abandoned.
There is a version of the EMF conversation that never gets past the first argument. Someone raises a concern. Someone else points to a safety limit, a study that found no effect, or the fact that we have all been living with wireless technology for decades without obvious catastrophe. The concern gets filed under anxiety or technophobia, and the conversation ends before it starts.
That version of the conversation is about framing, not science.
The framing matters because it determines what questions get asked, what gets measured, and what counts as evidence. If the framework says biology only responds to fields strong enough to heat tissue, then anything below that threshold gets declared safe by definition. Any study finding effects below it gets dismissed as a methodology error. That is not a scientific conclusion. It is a circular argument built into the measurement system itself. [If you want to understand exactly how those safety standards work and where they fall short, this breakdown covers it in detail: Why EMF Safety Standards Don't Measure What Actually Matters →]
Your body does not experience the electromagnetic environment as a SAR reading on a meter. It experiences it as conditions. Signals overlap, pulse, modulate, and interact in ways that shift moment to moment. The biological systems responsible for maintaining function inside that environment, your ion channels, your autonomic nervous system, your mitochondria, your neural timing networks, are sensitive to structure and pattern, not just magnitude. [For a full explanation of how biology responds to electromagnetic conditions as a signaling system: Your Body Is a Signaling System →]
On March 16, 2026, a thread went viral on X that appeared to turn this entire conversation on its head. A credentialed researcher had dug into the government's own $30 million cancer study on RF radiation and came back with a surprising conclusion: the animals exposed to RF radiation lived longer and were healthier than the ones that weren't. Four million people saw the TLDR. Very few saw what the data actually said.
Understanding what it actually said, and why the tools used to read it were the wrong ones for the job, opens a window into something much larger than one viral thread. It gets to the heart of why the EMF conversation has been so hard to have, why the research infrastructure built to study it was quietly shut down, and why the question of what the modern electromagnetic environment costs biology is one that still does not have an official answer.
On March 16, 2026 a thread went viral on X — Zane H Koch, voicing concern that his AirPods were frying his brain, then explained that he had dug into the data and was surprised by what he found. Not only were AirPods not frying his brain, the government's own cancer study showed RF radiation might actually be good for you. The TLDR at the end of the thread stated it cleanly:
Koch — a longevity and aging researcher with a bioinformatics PhD from UC San Diego, now on technical staff at Edison Scientific — had dug into the 2018 National Toxicology Program (NTP) study. The NTP is a program within the NIH; the 2018 study was the government's $30 million, decade-long rodent carcinogenicity program on cell phone radiation — one that found clear evidence linking RF exposure to cancer in rats. Contradicting those findings using the government's own data, Koch came back with the opposite conclusion: the study appeared to show not only that RF-exposed animals lived longer than the unexposed controls, but were healthier to boot. His explanation was hormesis — a biphasic dose-response model in which low-level stressors produce adaptive benefits rather than harm.
The timing couldn’t have been more perfect — Apple had announced the AirPods Max 2 that same day (Whether the algorithm served Koch's thread to millions because of it, or whether he simply asked the right question at the right moment, is a question only the algorithm can answer). The thread got 4 million views. The argument was simple and seductive: the government's own tumor study, when read carefully, exonerates your earbuds. You're fine.
Koch is exactly the kind of person who we train to examine this kind of data. He is trained in bioinformatics, giving him the statistical tools to go straight to the source. In this case though, that training is also what led him astray. He applied the right methods to the wrong kind of system. Studying the biological effects of electromagnetic fields is not like studying the effects of a drug, a toxin, or any system where cause and effect scale together in predictable ways. EMF biology doesn't follow those rules — and what makes it different needs to be understood before we get to what was actually killing the control rats.
There's also a simpler problem. Koch's analysis identified 28 apparent improvements against 5 harms in the exposed animals. When you run enough comparisons on the same dataset without pre-specifying your hypotheses, some will show positive results by chance alone. This is a statistical artifact that is so common it has a name: the multiple comparisons problem. The 28 improvements are not independent signals. They are the expected noise of an undisciplined search.
But even if you accept every one of those 28 improvements at face value, the TLDR leaves something out: the animals that "lived longer and were healthier" still developed cancerous tumors. The survival advantage had a specific explanation, and it had nothing to do with general health benefits from RF exposure.
A Brief History of Getting the Method Wrong on Purpose
Research into the biological effects of electromagnetism found its footing when Dr. Robert O. Becker set out to use bioelectricity to regenerate bone in non-union fractures. Becker and his lab assistant and physics mastermind Dr. Andrew Marino spent the better part of the 1970s and 1980s documenting those effects firsthand — and watching their findings ‘debunked’ by industry-financed reviews designed to bury them. Becker, twice nominated for the Nobel Prize, wasn't an anti-technology ideologue. He cam to the subject through his belief that we could use electromagnetic fields for healing. His bone regeneration research was initially dismissed by the medical establishment, but the FDA eventually approved it through a different door — work that gave rise to the PEMF devices used in orthopedic medicine today.
Becker and Marino later showed that the same electromagnetic principles capable of stimulating tissue repair were equally capable of causing harm when applied outside the controlled conditions of a therapeutic treatment. The wrong frequency, wrong modulation, wrong duration, wrong pulse rate, continuous exposure rather than targeted dosing — each a variable that could tip the same mechanism from healing to harm.
What they encountered in the research arena, while attempting to warn the public of the harms of high voltage power lines, was a systematic effort to design studies that would not find effects. The technique was as insidious as it was straightforward: apply linear statistical analysis to biological systems governed by nonlinear dynamics. Average across enough test subjects and the real effects washed out into noise. Declare the noise non-significant. Publish the null result. The approach produced decades of "inconclusive" findings — not because the biology was inconclusive, but because the wrong tool was applied to it consistently enough that the inconsistency became the official scientific record.
Becker told 60 Minutes, before he lost his funding, that it wasn't just harmful effects that were buried. Studies showing any biological effect from electromagnetic fields — positive, or negative — were suppressed. The threat wasn't findings that made industry look bad. It was findings that confirmed the fields were biologically active at all.
The Nonlinear Problem
Most people conflate complicated with complex. A simple system has few parts and behaves predictably — heat water to 100°C and it boils. A complicated system has thousands of parts, but those parts follow knowable rules. You can model it, engineer it, replicate it. A jet engine is complicated. Sending a rocket to the moon is complicated. Add enough data, enough computing power, and a complicated system becomes transparent. A complex system is categorically different. Its parts interact through nonlinear feedback loops, and the whole behaves in ways you cannot predict from studying the pieces. Small inputs produce disproportionate effects. The system is self-organizing, adaptive, and sensitive to initial conditions in ways that make detailed prediction practically impossible. Economies are complex. Ecosystems are complex. Biological systems are complex. Weather is complex. And electromagnetic field biology, it turns out, is complex. [This is also why measuring field strength alone misses the actual problem: The Problem Is Complexity, Not Power →]
Complex systems are governed by what mathematicians call chaos. Chaos is not randomness, but a counterintuitive kind of order in which tiny changes in initial conditions produce wildly different outcomes, with no smooth gradient between them, and no immediately obvious connection between input and output. James Gleick's 1987 book Chaos: Making a New Science brought this framework to public attention and was a finalist for both the National Book Award and the Pulitzer Prize. The science it described had been overturning assumptions across physics, economics, and meteorology for more than two decades by the time it reached a popular audience.
Edward Lorenz stumbled onto chaos theory in 1963 while running simulations to measure atmospheric circulations. He wanted to examine the results of one simulation in more detail, but upon re-running it he accidentally entered a starting value rounded to three decimal places instead of six. The results diverged completely from the original. It wasn't a change in degrees, or wind speed, or precipitation, it was an entirely different weather system. A difference of 0.000127 in a starting number had run through the nonlinear equations and resulted in a completely different outcome. This is what became known as the butterfly effect: the idea that a butterfly flapping its wings in Brazil could, through a chain of nonlinear amplifications, set off a storm over Texas.
It would be easy to mistake this for proof that the atmosphere is random, but it isn't. It is deterministic but exquisitely sensitive to initial conditions. This means that averaging across many measurements, as classical statistics does, destroys the actual information within the system. You don't get the middle of the distribution, you get noise. In order to predict a chaotic system with any certainty, you would need two things: an infinite number of sensors capturing every variable at every point in the system, and prior knowledge of the equation governing the system. If you miss one variable, or misread another, the prediction diverges. This is not an engineering problem that more computing power will eventually solve, it is a mathematical property of the system.
Benoit Mandelbrot arrived at the same framework while studying coastlines and financial markets. He found that natural systems generate infinite complexity from simple nonlinear rules — patterns that repeat at every scale, no matter how far you zoom in or zoom out. This self-similarity is what he called a fractal, and it never smooths out into the averages classical statistics expects to find. The Mandelbrot set, the fractal that became his signature image, while appearing almost like painting by Jackson Pollock, is actually a map of the boundary between order and chaos in a simple iterative equation.
The fractal shapes in these images are more than just mathematical curiosities. Take the Koch curve: start with a straight line, replace the middle third with two sides of an equilateral triangle, then repeat that operation on every resulting segment, infinitely. What begins as a line becomes a six-sided star, then something resembling a snowflake, its perimeter growing without bound while its area stays finite. No matter how closely you zoom in, the same structure repeats itself.
The Koch Snowflake. (It is not lost on me that this fractal shares a name, by coincidence, with the researcher whose thread started this conversation)
The Fractal Antenna
As it turns out, these shapes are also extremely efficient antenna designs. In 1988, a radio astronomer named Nathan Cohen heard Mandelbrot speak at a conference in Hungary and went home to build an antenna shaped like a fractal. It worked better than the conventional version, at a fraction of the size. Think of it like the difference between a flute and a French horn. A flute plays one note well — but to hit a lower note, you need a longer flute. A French horn coils yards of tubing into a compact circular shape, letting a musician play those deep, long-pipe notes without hauling a twelve-foot straight pipe around. Fractal antennas work the same way: they fold a long electrical component into a short physical space, allowing one tiny antenna to cover frequencies that would otherwise require an array of separate poles. Cohen patented the design in 1995. Within a decade, fractal antennas were inside every cell phone on earth, replacing the extendable antennas that used to telescope out of the top of handsets.
The irony is worthy noting:
The mathematics that describes why linear analysis fails on complex biological systems is the same mathematics that engineers applied to make cell phones smaller, more powerful, and capable of broadcasting across more frequency bands at once.
Mandelbrot's boundary between chaos and order is everywhere in this technology — etched into the antenna, expressed in the signal, and present in the biology on the receiving end. Our cells contain structured water arranged in crystalline domains with fractal-like organization.The collagen in our fascia forms fractal networks that repeat from the scale of whole muscle groups down through sheets of connective tissue, individual fibers, fibrils, and into the molecular structure of the collagen triple helix itself. Our vascular tree branches fractally. So does our nervous system. We are, in the most literal sense, a fractal antenna — and a far more sophisticated one than anything etched on a circuit board. If a simple fractal pattern on a circuit board can receive signals across the electromagnetic spectrum, what might the extraordinarily complex fractal architecture of living tissue be capable of receiving? [This is also part of why man-made fields interact with biology differently than natural ones — the structure of the signal matters as much as its strength: Why Man-Made EMF Is Different From Natural EMF →]
The Experiment
In that biology, relevant variables aren't just dose and duration, they include frequency, modulation pattern, pulse rate, coherence, and the specific geometry of the signal relative to the biological structure it interacts with. Change one of those parameters and you can cross from no effect to significant effect, or from stimulation to inhibition, without passing through a gradient warning of the new effect. In this kind of system, dose is one variable among several, and not necessarily the decisive one. The system is deterministic, and governed by specific rules, but those rules don't show up in a SAR reading, or a hormesis model. There's no biphasic dose-response curve or predictable crossover. The response can be present at one set of parameters and gone at a nearly identical one, with nothing in between to warn you.
Marino built a system to prove this, but he didn't do it alone. Working with mathematicians specializing in nonlinear dynamics, who had applied the same tools used to study chaotic systems in physics and economics, he designed experiments that could detect structured patterns. Human subjects sat in a controlled environment while weak electromagnetic fields — 1 to 5 microtesla, the kind of field strength you'd encounter near ordinary electrical wiring — were switched on and off in sequences the subjects were not able to monitor, and couldn’t perceive. Electrodes recorded the subjects' EEG continuously, capturing the electrical activity of the brain across those exposure windows. When the researchers averaged those EEG readings across trials the way it is done with linear analysis no effect was found. Exactly the kind of null result that had been filling the literature for decades.
Then they applied a technique called Recurrence Quantification Analysis, a mathematical tool developed for detecting structure in systems that look random on the surface. The core idea is this: plot every measurable variable of the system against every other variable, at every moment in time, building a kind of fingerprint of how the system moves through all its possible states — and look for the moments when the trajectory returns to where it has visited before. In a purely random system, the trajectory just wanders. In a deterministic system — even a chaotic one — it does return, tracing recognizable patterns at recognizable intervals. RQA quantifies those returns: how often they occur, how long they persist, how structured the pattern is. White noise produces a scatter of isolated dots. A deterministic system produces diagonal lines. The length and density of those lines is the signature of underlying order.
When Marino's team ran the EEG data through RQA, the scatter resolved into lines. The brain responses were not noise, they were reproducible, field-dependent, and governed by the parameters of the signal rather than its intensity alone. What Marino and his collaborators could not do — and did not claim — was predict exactly when or how a given exposure to a given individual would produce a given response. That is the nature of chaotic systems. The Lorenz equations can tell you the atmosphere is a deterministic system without letting you pinpoint next month's weather. What Marino proved was the equivalent: that the biological response to weak electromagnetic fields is not noise. It is a deterministic system. The rules exist. They are just not the rules the regulatory framework has been measuring.
This is why It is not paradoxical that the NTP tumor signal peaked at the lowest dose rather than the highest.
Independent Confirmation from a Different Direction
The same nonlinear logic shows up in regenerative biology, and undeniable evidence of its existence arrived from a study that had nothing to do with cell phone safety. Planarian flatworms are the go-to model organism in regenerative biology because they regenerate from fragments — slice one in half and both halves grow back into full planaria. They are the ideal system for studying stem cell behavior, wound healing, and tissue growth. Nobody funds planarian research to study the biological effects of cell phones, they fund it to study regeneration. The electromagnetic findings are a byproduct of asking a different question, which means no one had a reason to design the experiment to fail, and thus the studies get funded.
Wendy Beane, a biologist at Western Michigan University, exposed planarians to weak static magnetic fields across a range of field strengths below 1 milliTesla. The results did not fit a dose-response curve — and they did not fit a hormetic curve either. At 200 μT, the fields suppressed reactive oxygen species (ROS) formation – which can lead to oxidative stress and cellular and DNA damage – and blocked stem cell proliferation, inhibiting regeneration. At 500 μT, the exposure produced the opposite outcome: elevated ROS, increased stem cell proliferation, and accelerated tissue growth. A 2023 follow-up identified superoxide as the reactive oxygen species being modulated, and provided the explanation: weak magnetic fields alter electron spin states via the radical pair mechanism, switching the system between outcomes in a binary “on/off” pattern, rather than in a graded fashion.
The same superoxide pathway sits at the center of the mechanistic framework we'll examine in the next section. The two research programs — planarian regeneration biology and RF carcinogenicity — arrived at the same reactive oxygen molecule from opposite ends of biology, with no coordination between them. This is what independent replication looks like when the effect is real.
The Control Group Problem
The NTP's Technical Report is explicit about what happened to the unexposed animals: "Survival of all exposed male groups was significantly greater than that of the sham control group due to the effect of chronic progressive nephropathy (CPN) in the kidney of sham control males." The controls died earlier because they developed more severe age-related kidney disease. The RF-exposed males had the same incidence of CPN but much lower severity — fewer deaths from kidney failure and the downstream organ complications it causes. That is the effect the viral thread was reading as a longevity benefit.
In response to the thread, John Coates of RF Safe pointed to a brand new, FDA-approved device for treating advanced hepatocellular carcinoma, as evidence for the mechanism that explains how the kidney protection happened. The device also explains why Koch’s analysis doesn't support the hormesis claim.
Coates knows this territory intimately, both as a researcher and as someone who has paid its costs personally. Growing up on a military air base in Virginia in the 1970s, he had his left kidney removed after developing cancer as a child. He connects this condition to exposure to the microwave radar systems his father, a decorated Navy aviator, worked around. In July 1995, he held his newborn daughter, Angel Leigh, for a few hours before she died from anencephaly, a catastrophic neural tube defect. Two years later, a 1997 study showed that pulsed electromagnetic fields could induce the identical defect in chicken embryos. He founded RF Safe in 1998 as a promise – a one man mission to transform how we build and use wireless technology. I wanted to share that background before going over his analysis. Coates is not unusual in the EMF research community — many of the people who have stuck their necks out, faced professional ridicule and often even career suicide, and kept pushing have done so because they or someone they loved paid an immeasurable price. That's a different kind of credential – that is skin in the game.
TheraBionic is a non-invasive liver cancer treatment. It delivers radiofrequency at roughly 1,000 times lower power than a phone, with tumor-specific wave form properties designed to target liver cancer cells while leaving surrounding tissue unaffected. It works because finely calibrated frequencies and modulations interact with specific tissue in tissue-specific ways. That is the point — and it is the exact opposite of what the viral thread claimed. Tissue specificity at therapeutic precision is not evidence that untargeted pulsed RF radiating into the body at phone power levels will lead to a general adaptive benefit.
None of this means therapeutic EMF is without risk. Robert Becker himself, whose work on bioelectricity inspired the PEMF devices now used in orthopedic medicine, warned that stimulating cell proliferation and tissue regeneration with electromagnetic fields carried an inherent danger: you cannot always control the difference between healing tissue and cancer — both are growth, one is directed, and one is not. The precision that makes TheraBionic work is what makes it less dangerous than ambient exposure, but not necessarily safe in any absolute sense.
We are doing our best to navigate a nonlinear system we do not fully understand. Precision is necessary but not sufficient. What TheraBionic represents is careful empiricism inside a complex system, not mastery of it. They ran the trials and observed what worked. We accept the remaining uncertainty because the alternative for a patient with advanced liver cancer is often interventions with a much higher risk profile. That calculus is easily defensible. Deploying the same class of energy — less precisely, at higher power, continuously, across billions of people who did not consent to the experiment — is a different calculus altogether.
What the NTP Study Actually Found
Set aside what was killing the controls, and the NTP findings are not ambiguous. The program found clear evidence of an association with malignant schwannomas in the hearts of male rats — "clear evidence" being the NTP's highest rating on a four-level evidentiary scale. Some evidence of malignant gliomas in the brain. Some evidence of pheochromocytoma in the adrenal glands. DNA damage in the frontal cortex of male mice, the blood cells of female mice, and the hippocampus of male rats. These findings were reported after NTP statisticians corrected for the survival difference — the tumors are present regardless of how long the controls lived.
The most dangerous tumor findings — malignant heart schwannomas and brain gliomas — peaked at the lowest exposure level tested, 1.5 W/kg, not at the highest 6 W/kg. The maximum SAR for use of a current iPhone reaches 1.60 W/kg.
The dose that produced the clearest cancer signal in the study is the dose closest to what people are actually experiencing.
Why the Low-Dose Peak Isn't Surprising
The detail – the unexpected finding of more tumors at lower doses - is confounding to most people. Classical toxicology has no framework for that, but the voltage-gated ion channel literature does.
Dimitris Panagopoulos formalized the mechanism in a 2025 paper in Frontiers in Public Health. Pulsed RF signals carry an extremely-low-frequency component embedded in the modulation structure as an artifcat of the on/off pulsing patterns that encode the signal. The geometry of that ELF component aligns with the physical structure of voltage-gated ion channels and cell membranes, forcing the ions near those channels to oscillate. The channels are normally gated by much stronger voltage changes (3×10⁶ and 10⁷ V/m), but the ion forced oscillation mechanism shows they can also respond to polarized, coherent, slow-varying fields that are exponentially weaker (as weak as 10⁻⁵ V/m). The result is irregular channel opening, disrupted ion concentrations inside the cell, and a surge of ROS overproduction downstream. When VGCC blockers are introduced experimentally, the effects disappear — which confirms that the pathway is specific, and not a general stress response.
SAR, the specific absorption rate used as the regulatory metric, is a proxy for a thermal mechanism. It measures how much energy is absorbed as heat. It says nothing about pulse rate, modulation pattern, or the coherence structure that actually drives the ion forced oscillations. Calling for revised standards is warranted because the current limits came from a framework that doesn't measure the right variables at all, and because total load across all tissues still accumulates with exposure time and intensity. The precautionary principle holds even without a complete mechanistic explanation: the existing data show harm at the exposure levels the NTP study used — levels roughly equivalent to what people currently experience. Minimize the noise until the science can characterize the actual parameter specific dose-response.
What that means in practice is not simply a lower SAR number stamped on a box. It means engineering devices differently. Phones that drop transmission power when a strong signal makes high power unnecessary. Radios that turn off in standby mode, rather than pulsing beacons at regular intervals. Antennas that direct radiation away from the body rather than into it. No simultaneous WiFi and cellular transmission when one will do. Bluetooth off by default. All radios off when a wired connection is available. The goal is reducing the total electromagnetic noise the body is subject to — frequency, pulse rate, modulation pattern, coherence and duration — not just trimming a single, arbitrary metric.
The Melnick Reanalysis
Ronald Melnick led the design of the original NTP study. On March 14, 2026, he and Joel Moskowitz published a paper in Environmental Health on behalf of the International Commission on the Biological Effects of Electromagnetic Fields. The methodology was deliberately orthodox. They used the standard EPA benchmark-dose modeling, the same risk-assessment framework used for every other environmental carcinogen. No exotic methods. No nonlinear modeling. The most conventional tool in the regulatory arsenal, applied to two independent datasets. The result was the same.
Let’s pause here. The entire essay has argued that linear analysis is the wrong tool for EMF biology — and it is. But Melnick and Moskowitz used it deliberately, using the most conservative assumptions available, on data the industry has cited for decades. If even that blunt instrument finds limits 200 times too high, the question of what a nonlinear analysis would find is one the regulatory framework has spent fifty years making sure nobody asks.
The first data set was the NTP study itself. The second was the Ramazzini Institute study, published the same year, 2018, by researchers in Bologna, Italy. The Ramazzini team ran a lifetime exposure study on rats using a GSM signal at 1.8 GHz. They found the same heart schwannomas in male rats that the NTP had found, at whole-body exposure levels of 0.1 W/kg — roughly 100 times lower than the lowest NTP dose.
Applied to both datasets, the result showed that current FCC and ICNIRP whole-body public exposure limits are at least 200 times too high to protect against a 1-in-100,000 lifetime cancer risk, assuming 8 hours of daily exposure. For male reproductive harm — decreased sperm count, sperm vitality, and testosterone — the current limits are 8 to 24 times too high. The protective exposure level for cancer risk falls between 0.8 and 5 mW/kg. The current public limit is 80 mW/kg — milliwatts per kilogram of tissue, a measure of how much electromagnetic power is absorbed per unit of body mass.
The FCC last updated its RF exposure limits in 1996. Those limits were derived from a handful of studies done in the 1980s, calibrated entirely around the threshold for thermal tissue damage. Melnick helped generate the data that most directly challenges whether that threshold has anything to do with cancer.
The Study That Was Halted
That the 2018 NTP findings reached the public at all was not inevitable. The study was commissioned in 1999 by the FDA and nominally designed to test what the regulatory establishment expected to confirm: that cell phone radiation at non-thermal levels could not cause adverse health effects. John Bucher, the NTP senior scientist who co-directed the study, stated publicly that he expected the results to show no association between RF radiation and cancer. An anonymous senior government radiation official told reporters at the time: "Everyone expected this study to be negative." When the results came back showing clear evidence of cardiac schwannomas and some evidence of brain gliomas, the findings surprised the scientists running the experiment. The FDA — the same agency that had nominated the study — responded to its own commissioned research by declaring that the findings "should not be applied to human cell phone usage" and that existing safety limits "remain acceptable."
Because the 2018 study had examined only 2G and 3G frequencies — the technology of the late 1990s when the program was designed — a follow-up was planned to examine newer modulations. The research infrastructure had been built and the exposure systems had been validated. By 2023, 5G was deploying at scale. The follow-up would have been the first government research examining what the technology people were actually using today did to biological tissue under controlled conditions.
In January 2024, NIEHS released a fact sheet officially ending the program. The stated rationale: the small-scale follow-up exposure system had proven more technically challenging and resource-intensive than anticipated, and was designed for 2G and 3G frequencies — "not representative of newer technologies such as 4G/4G-LTE, or 5G." The program studying whether cell phone radiation harms human biology was discontinued because the technology had advanced faster than the science could follow. The proposed response was not to fund research that could follow.
Devra Davis, a former senior adviser at HHS and founder of the Environmental Health Trust, called it what it was in an opinion piece in The Hill: a glaring abdication of responsibility. The NTP's own 2023 fact sheet had pledged follow-up studies to better understand the long-term animal findings. Results from the small-scale exposure system that had been developed and validated were neither published nor publicly shared before the program ended. The public infrastructure built to research what wireless radiation does to biological tissue was retired without fanfare — conveniently timed to just before the wireless industry prepared to roll-out the next generation of that infrastructure.
What Replaced It
Tensions around wireless are emerging inside the current administration, and they are pulling in opposite directions simultaneously.
In December 2025, the White House released a Presidential Memorandum directing immediate clearance of reserved spectrum for full-power commercial 6G use. Federal agencies currently using that spectrum — military, weather, scientific — have 12 months to develop plans for vacating it so the bandwidth can be handed to commercial carriers. Studies on two additional bands are underway for reallocation as well. The administration's legislative framework targets 800 MHz of new licensed spectrum to be delivered by 2034.
In January 2026, under the same administration, HHS Secretary RFK Jr. announced a new federal study on the health effects of electromagnetic radiation from cellphones. The FDA removed web pages that had for years stated the "weight of scientific evidence has not linked exposure to radiofrequency energy from cellphone use with any health problems." HHS spokesman Andrew Nixon described the removed pages as containing "old conclusions." Kennedy told USA Today he was "very concerned" about electromagnetic radiation, pointing out that European exposure regulations are an order of magnitude lower than what the US permits.
The FCC, meanwhile, is greenlighting spectrum auctions and proposing to further preempt state and local permitting rules to accelerate wireless infrastructure deployment. Its website still states that "no scientific evidence establishes a causal link between wireless device use and cancer or other illnesses." It has not complied with a 2021 D.C. Circuit court order directing it to explain how its current limits adequately protect human health.
There is no current federal research program examining what 5G does to biological tissue, and there is no announced plan to create one. The infrastructure is being built on the assumption that the question has been answered — by agencies whose exposure guidelines were written when the Soviet Union still existed, calibrated around whether a signal heats your skin. [New to EMF and want to understand the basics before going deeper? Start here: What Is EMF? →]
Zane Koch posted simple survival curves in his X thread. He didn't post the kidney pathology tables, the tumor incidence data, or the EPA risk numbers. His TLDR: "the mice/rats were bathed in radiofrequency radiation at doses 10-100x higher than AirPods, 9 hrs/day, for 2 years. and. lived. longer. and. were. healthier."
The two signals in this study are not in conflict. The survival signal comes from how the controls died. The cancer signal comes from what the exposure did to specific tissue in lab rats. Two different measurements of two different things, reported in the same dataset, and read as one.
The animals with the best kidneys got heart cancer. That's the study.
In Conclusion
The NTP study was not designed to produce the findings it produced. The researchers running it expected a null result. When the data came back showing clear evidence of cardiac schwannomas and evidence of brain gliomas, the agencies responsible for acting on those findings questioned the study's relevance to human exposure, funded no follow-up research capable of resolving the uncertainty, and eventually shut the program down entirely. The regulatory limits governing public exposure to wireless radiation have not been meaningfully updated in three decades. The reanalysis published in 2026 by the scientist who designed the original study found those limits are at least 200 times too high to protect against a meaningful cancer risk threshold.
None of that makes the answer simple. Biology is genuinely complex. The variables that determine biological response go far beyond what any single study can capture. What is clear is that the framework the regulatory system has relied on: power, heat, and a single threshold, is not measuring what biology actually responds to.
Biology responds to conditions. Signal structure, pulse patterns, coherence, the cumulative density of overlapping fields from every device operating in a shared space. It responds to the electromagnetic environment as a whole, and the body compensates when those conditions become unstable. That compensation has a cost. It shows up quietly, across systems, long before anything visibly fails.
Aires was built around this understanding. The starting point was not that wireless technology is dangerous. It was that the electromagnetic environment has become genuinely complex in ways biology was not shaped to navigate, and that the right response to complexity is better design. The fractal geometry at the core of every Aires device is a structural answer to a structural problem. When fields are chaotic and unpredictable, introducing coherent organization into that environment reduces the variability that biological systems have to continuously work against. Not by blocking signals. Not by eliminating technology. By improving the conditions biology is operating inside.
The conversation this blog traces, from a viral thread to a suppressed government study to a reanalysis that deserves far more attention than it has received, is the same conversation Aires has been part of for years. That science is catching up to the problem. The electromagnetic environment is not going to become simpler on its own. Clarity has to be built into the conditions.
That is exactly what Aires is designed to do.
Note: The viral thread received a prominent mention on the DarkHorse Podcast with Bret Weinstein and Heather Heying as I was about to publish. Expect follow-up on their coverage soon.
https://www.youtube.com/watch?v=ZoZIte3T0Ec
References
-
@zanehkoch, X, March 15, 2026 — https://x.com/zanehkoch/status/2033400775762964601
-
Apple Inc., "Apple Introduces AirPods Max 2," March 16, 2026 — https://www.apple.com/newsroom/2026/03/apple-introduces-airpods-max-2-powered-by-h2/
-
@ze_rusty, X, March 16, 2026 — https://x.com/ze_rusty/status/2034160805370060934
-
Becker, R.O. & Selden, G. — The Body Electric: Electromagnetism and the Foundation of Life, HarperCollins — https://www.harpercollins.com/products/the-body-electric-robert-beckergary-selden?variant=32117253832738
-
Robert O. Becker Archive — "60 Minutes: Project Seafarer," February 13, 1977 — https://robertobecker.net/later-life/60-minutes/
-
Litovitz et al. — "Bioeffects induced by exposure to extremely low-frequency electromagnetic fields," PubMed, 1997 — https://pubmed.ncbi.nlm.nih.gov/9261540/
-
Marino, Nilsen, Frilot — "Consistent magnetic-field induced dynamical changes in rabbit brain activity detected by recurrence quantification analysis," Brain Research, 2002 — https://www.sciencedirect.com/science/article/pii/S0006899302040933
-
Marino et al. — Brain Research, 2002 (second RQA paper) — https://www.sciencedirect.com/science/article/abs/pii/S0006899302031773
-
Marino, Nilsen, Frilot — "Nonlinear changes in brain electrical activity due to cell-phone radiation," Bioelectromagnetics, 2003 — https://andrewamarino.com/ee_electroencephalogram/
-
Carrubba, Frilot, Chesson, Webber, Zbilut, Marino — "Magnetosensory evoked potentials: consistent nonlinear phenomena," Neuroscience Research, 2008 — https://www.sciencedirect.com/science/article/abs/pii/S0168010207017956
-
Marino, A.A. — Full publication archive — https://andrewamarino.com
-
National Toxicology Program — TR-595, Cell Phone RFR Studies in Rats, 2018 — https://ntp.niehs.nih.gov/go/tr595
-
National Toxicology Program — TR-596, Cell Phone RFR Studies in Mice, 2018 — https://ntp.niehs.nih.gov/go/tr596
-
Scientific American — "Major Cell Phone Radiation Study Finds Cancer Link," May 27, 2016 — https://www.scientificamerican.com/article/major-cell-phone-radiation-study-finds-cancer-link/
-
The Nation — "How Big Wireless Made Us Think That Cell Phones Are Safe," March 29, 2018 — https://www.thenation.com/article/archive/how-big-wireless-made-us-think-that-cell-phones-are-safe/
-
Falcioni et al. — "Brain and heart tumors in Sprague-Dawley rats exposed to mobile phone radiofrequency field," Environmental Research, 2018 — https://doi.org/10.1016/j.envres.2018.01.037
-
Van Huizen & Beane — "Weak magnetic fields alter stem cell–mediated growth," Science Advances, 2019 — https://www.science.org/doi/10.1126/sciadv.aau7201
-
Smith-Roe et al. — "Evaluation of the genotoxicity of cell phone radiofrequency radiation," Environmental and Molecular Mutagenesis, 2019 — https://doi.org/10.1002/em.22343
-
U.S. FDA — TheraBionic P1, HDE H220001, approved September 26, 2023 — https://www.fda.gov/medical-devices/recently-approved-devices/therabionic-p1-h220001
-
NIEHS — NTP Cell Phone RFR program closure statement — https://ntp.niehs.nih.gov/research/topics/cellphones
-
NIEHS — Cell Phone RFR Studies fact sheet, January 2024 — https://www.niehs.nih.gov/sites/default/files/NTP_cell_phone_factsheet_jan_2024_508.pdf
-
Davis, D. — "Why did NIH abruptly halt research on the harms of cell phone radiation?" The Hill, February 1, 2024 — https://thehill.com/opinion/technology/4437988-why-did-nih-abruptly-halt-research-on-the-harms-of-cell-phone-radiation/
-
Kinsey & Beane — "Weak magnetic fields modulate superoxide to control planarian regeneration," Frontiers in Physics, 2023 — https://www.frontiersin.org/journals/physics/articles/10.3389/fphy.2022.1086809/full
-
Pall, M.L. — "Electromagnetic fields act via activation of voltage-gated calcium channels," Journal of Cellular and Molecular Medicine, 2013 — https://doi.org/10.1111/jcmm.12088
-
Panagopoulos, Yakymenko, De Iuliis, Chrousos — "A comprehensive mechanism of biological and health effects of anthropogenic ELF and wireless communication electromagnetic fields," Frontiers in Public Health, 2025 — https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2025.1585441/full
-
Environmental Health Trust et al. v. FCC — D.C. Circuit, 2021, petitioners' opening brief — https://ehsciences.org/wp-content/uploads/2025/10/EHT-et-al-FCC-Opening-Brief-11122020-.pdf
-
FCC — RF Safety FAQ — https://www.fcc.gov/engineering-technology/electromagnetic-compatibility-division/radio-frequency-safety/faq/rf-safety
-
RF Safe / John Coates, X, March 2026 — https://x.com/rfsafe/status/2034165700999127380
-
The White House — "Fact Sheet: President Trump Takes Action to Win the 6G Race," December 19, 2025 — https://www.whitehouse.gov/fact-sheets/2025/12/fact-sheet-president-donald-j-trump-takes-action-to-win-the-6g-race/
-
Congress.gov — H.R. 1, Working Families Tax Cuts Act (Big Beautiful Bill), 119th Congress — https://www.congress.gov/bill/119th-congress/house-bill/1
-
HHS / Andrew Nixon statement, January 15, 2026; Reuters — https://www.reuters.com/legal/litigation/us-health-department-launch-study-cellphone-radiation-2026-01-15/
-
Kennedy, R.F. Jr. — USA Today interview, January 16, 2026, reported via Fox News Digital — https://www.foxnews.com/health/cellphone-health-risks-focus-new-government-study-very-concerned
-
Broadband Breakfast — "RFK Jr. Very Concerned About Cellphone Radiation, Launches Safety Study," January 20, 2026 — https://broadbandbreakfast.com/rfk-jr-very-concerned-about-cellphone-radiation-launches-safety-study/
-
Melnick, Moskowitz, ICBE-EMF — "Exposure limits to radiofrequency EMF do not account for cancer risk or reproductive toxicity," Environmental Health, 2026 — https://link.springer.com/article/10.1186/s12940-026-01288-6
-
ICBE-EMF — Letter to U.S. Congress, January 14–15, 2026 — https://icbe-emf.org/u-s-and-international-scientists-urge-congress-to-question-fcc-over-outdated-cell-tower-radiation-safety-limits/
- Zhang et al. — "Brain Accumulation of Airborne Magnetite Nanoparticles under Earphone/Smartphone-Embedded Magnetic Fields Triggers Neurotoxicity," ACS Nano, 2026 — https://pubmed.ncbi.nlm.nih.gov/41774862/