As of mid‑January 2026, Reuters reports that HHS is launching a study on cellphone radiation, and that FDA “took down old webpages saying cellphones are not dangerous” while HHS undertakes that work.
That is a critical framing: pulling categorical webpages is not the same thing as declaring a confirmed hazard. It is a risk‑communication and scientific‑integrity move: stop presenting closed‑case certainty (“not dangerous”) when the evidence base is complex, evolving, and includes credible signals that do not support blanket reassurance.
2) Why “assurances of safety” are the wrong level of certainty for the evidence base
A. FDA’s prior “no danger” language was strong enough to be treated as a definitive safety claim
The FDA’s “Cell Phones” page still contains language such as “The weight of scientific evidence has not linked cell phones with any health problems” and “The scientific evidence does not show a danger to any users of cell phones from RF exposure, including children and teenagers.”
Even more importantly, the D.C. Circuit documented that FCC relied on those exact kinds of FDA statements—including the “no health problems” phrasing and the “no danger…including children and teenagers” phrasing—when deciding to keep long‑standing RF exposure policies.
B. A federal appeals court explicitly criticized those FDA statements as “conclusory”
In Environmental Health Trust v. FCC (D.C. Cir. 2021), the court held the FCC’s order was arbitrary/capricious in relevant part because it relied on conclusory FDA statements that did not explain the factual bases for dismissing evidence in the record.
This matters for HHS/FDA communications even before you argue biology:
If a federal court has already highlighted that prior FDA safety‑sounding statements were effectively assertions without an articulated reasoning chain, there is an institutional scientific‑integrity argument for taking them down and rebuilding the public‑facing posture on a more transparent, evidence‑graded foundation.
3) What the weight of evidence actually supports (and why that undercuts categorical reassurance)
If the public message is “no danger,” then the evidentiary bar should be: no credible signals across human, animal, and mechanistic domains that would make that claim overconfident. The current record fails that bar.
A. Human epidemiology is not a clean “all clear,” especially at the high‑use end
INTERPHONE (IARC-coordinated) is often summarized as “no link,” but the IARC press release makes two points that directly undermine blanket reassurance:
-
In the highest decile of cumulative call time (≥1,640 hours), the odds ratio for glioma was 1.40 (95% CI 1.03–1.89).
-
That “heavy use” threshold corresponds to about a half‑hour per day over 10 years, and IARC explicitly notes that most participants were not heavy by today’s standards, with youth now commonly at an hour or more per day (even accounting for technology changes).
INTERPHONE alone does not “prove” causation (IARC emphasizes bias/error limitations), but it does support a narrower, defensible statement: the highest‑use tail showed signals that make a categorical “no danger” message too confident.
B. WHO‑commissioned human evidence reviews lean reassuring—but they don’t justify “no danger” phrasing
A WHO‑commissioned systematic review/meta‑analysis of human observational studies (Karipidis et al., Environment International, 2024) reported meta‑estimates around null for multiple tumor types (e.g., glioma mRR ≈ 1.01) and concluded exposure from mobile phone use likely does not increase brain cancer risk.
That is real evidence on the “reassuring” side. But even if you accept that conclusion, it still does not warrant a categorical “no danger to any users” claim, because:
-
“Likely does not increase risk” is not “risk is zero,” and it is not “no danger for any user,” especially given exposure heterogeneity, latency, and evolving usage patterns.
-
The WHO review is explicitly within a framework that continues to produce multiple SR‑MAs and acknowledges ongoing assessment needs; it does not close the scientific question forever.
So the intellectually honest endpoint is: public messaging should reflect uncertainty and the boundaries of inference, not blanket assurance.
C. Animal carcinogenicity evidence is unusually central here—and it cuts against blanket reassurance
Two independent, large rodent programs reported tumor findings that are highly salient to risk communication:
-
U.S. National Toxicology Program (NTP) concluded a link between RF exposure and tumors in male rats—famously “clear evidence” for malignant schwannomas of the heart (male rats), with additional evidence categories for other findings.
-
Ramazzini Institute (Falcioni et al., 2018) reported increased heart schwannomas in male rats at exposures intended to mimic far‑field/base‑station conditions (with reported SAR estimates in the study).
Then, critically, a WHO‑commissioned systematic review of animal cancer studies (Mevissen et al., Environment International, 2025) judged the certainty of evidence as high for malignant heart schwannomas in male rats and high for glioma, based on its grading approach.
Even if one disputes how the certainty grading was performed, Germany’s Federal Office for Radiation Protection (BfS) spotlight underscores the controversy itself: it describes the review and simultaneously critiques the methods and the certainty interpretations, while still recognizing that NTP in particular shows indications at high exposures.
From a policy standpoint, that combination (“two major animal programs + a WHO‑commissioned animal SR”) is more than enough to invalidate a categorical “not dangerous” public posture.
D. Mechanistic evidence provides plausibility at non‑thermal levels (again: plausibility, not proof)
Mechanistic reviews argue for pathways involving oxidative stress and DNA damage:
-
Yakymenko et al. review oxidative effects and oxidative damage endpoints in the RF literature.
-
Panagopoulos et al. (Frontiers, 2025) presents a consolidated mechanistic model in which EMF exposure can contribute to oxidative stress and downstream biological effects.
Mechanistic literature is legitimately debated on quality control, exposure characterization, and publication bias. But for communications policy, the key point is narrower: a plausible biological pathway has been articulated in peer‑reviewed literature; therefore, categorical “no danger” language is not scientifically conservative.
E. Population trends can’t prove causation, but they do argue against “closed case” messaging
Denmark’s official cancer registry report notes that brain/CNS tumor incidence rates rose from 2021 to 2023 (men 27.4 → 32.5, women 35.7 → 42.0, age‑standardized per 100,000).
No competent analyst should claim this proves RF causation; the report itself frames trends within broader epidemiologic context. But it is still incompatible with a public posture that implies “no health problems” in a way that would discourage vigilance and research.
4) The strongest “immediate removal” argument is fundamentally about scientific integrity and public risk communication
You do not need a “silver bullet” study to justify pulling categorical assurances. You need to show that the assurance exceeds what the evidence can support.
A scientifically defensible, immediate action is:
-
Remove or suspend web copy that reads as definitive safety certification (“not dangerous,” “no danger to any users”), because the evidence base includes:
-
a high‑exposure signal in a flagship human case‑control dataset
-
convergent animal tumor findings
-
a WHO‑commissioned animal SR asserting high certainty for key tumor types (with active methodological dispute)
-
plausible mechanisms in the literature
-
and a judicial record that prior FDA “totality” statements were treated as conclusory when used as a regulatory backstop
-
In other words, the case for removal is not that harm is proven; it is that “assured safety” is not proven—and official language should not imply it is.
5) A practical “replacement” posture that is consistent with the evidence (and avoids misleading the public)
If the goal is to stop misleading reassurance without overstating harm, the best practice is to move from categorical statements to graded, evidence‑bounded language such as:
-
Human evidence overall: “Current human observational evidence does not show a clear increased risk for typical use patterns, but limitations remain, including uncertainty at the highest cumulative exposures.”
-
Animal evidence: “Well‑conducted animal studies have reported tumor findings under certain exposure conditions; the relevance to human exposures continues to be evaluated.”
-
Mechanisms: “Research has proposed non‑thermal biological pathways (e.g., oxidative stress) that are under active study.”
-
Practical exposure‑reduction: “If you want to reduce exposure, use speakerphone or a wired headset, text more, and avoid carrying an active phone directly against the body.”
That posture preserves public trust better than either extreme (blanket reassurance vs. definitive alarm).
6) Bottom line
The best “totality of evidence” argument for immediate removal of FDA safety‑assurance language is:
-
The language went beyond what the evidence can support, and it has already been treated as a conclusory safety backstop in a major RF policy record.
-
The modern evidence base contains credible signals and graded conclusions (especially from animal carcinogenicity evidence) that make categorical reassurance scientifically unjustified.
-
Removing those statements while commissioning updated review is a science‑forward, uncertainty‑honest move, not an anti‑science move—provided replacement messaging is transparent and evidence‑graded.