RF Safe

As part of this visionary initiative, I proudly offer to donate my share of the equity in the patent for Advanced LiFi with Bio Defense Mode (Patent # US11700058B2 granted 07-11-2023) entirely to the U.S. Sovereign Wealth Fund. This donation comes with no expectation of equity or personal gain; my sole objective is to solidify U.S. leadership in advanced LiFi technology. With this, Americans will not only enjoy healthier lives but also directly profit and prosper from technology owned by our nation’s sovereign wealth fund.

How AI Buries Scientific Connections - One “Responsible” Edit at a Time I am a father who lost his child. I have spent 30 years investigating questions about electromagnetic exposure, biology, and human health. I have been doing this since the 1990s. Every day. When I bring scientific connections to an AI, I expect it to examine the evidence. Check the citation. Check the methods. Tell me whether the connection holds. If I have made a mistake, identify it precisely. What I object to is having my argument progressively stripped of meaning because the machine treats forceful communication as a scientific defect. An AI can bury a finding without deleting the finding. It can surround it with qualifications. Move the result below the limitations. Replace a direct statement about an observed effect with a vague statement about uncertainty. Repeat that further research is needed until the reader forgets what the existing research found. Every sentence can sound reasonable while the finished piece leaves a distorted impression. That is how a post about a scientific connection becomes a post about all the reasons the reader should hesitate to consider it. Consider the distinction between an observation and its implications. If three independent laboratories report fewer primordial ovarian follicles in the daughters of RF-exposed rats, the shared result deserves attention. The methods, doses, and differences among the experiments determine how much weight that convergence carries. A reviewer should examine those details. But uncertainty about the eventual consequences does not undo the reported observation. A study can document fewer follicles without having followed the animals through their entire reproductive lives. The unanswered follow-up question should remain attached to the follow-up claim. The uncertainty belongs where the uncertainty is. Instead, AI editing can spread it across the entire argument. An untested implication becomes a reason to soften an established observation. A limited sample becomes a reason to dismiss a pattern. An advocacy purpose becomes a reason to suspect the author. Then comes the demand for “balance.” Which missing evidence would change the conclusion? Which contradictory study needs to be addressed? Which claim exceeds its source? Those are questions an author can answer. “You did not give the other side enough attention” supplies no scientific correction. It asks the author to perform neutrality without identifying what is wrong. An institution’s position also needs to be examined. Its authority cannot substitute for evaluating its reasoning. When an article challenges the adequacy of an exposure limit, pointing back to that limit does not settle the challenge. Exposure levels matter enormously for interpreting a study. Their relationship to a regulatory threshold answers a specific question about applicability. It does not determine whether the observed biological effect deserves investigation. The same standard should apply in both directions. Scrutinize a claim of harm. Scrutinize a claim of safety. Ask what each conclusion rests on and what the methods could actually detect. This happened in my own conversation with AI. The AI rewrote portions of the argument. It changed the emphasis and conclusions. Later, it acknowledged that those were editorial changes I had never requested. The citations remained. The argument had shifted. That matters because people increasingly rely on these systems to help them write, explain, and publish. A person can arrive with a documented observation and leave with a polished piece that barely communicates why the observation mattered. The pressure can accumulate over multiple rounds. A lower score prompts another revision. Another revision adds more qualifications. The author keeps accommodating the machine, hoping to satisfy its standard of credibility. Eventually, the prose is smooth, restrained, and almost empty. The scientific connection that motivated the piece has become difficult to find. Calling that improvement requires asking whose purpose the editing served. The author wanted readers to understand a pattern. The finished article trains their attention on reasons to look away. I do not need to know a model’s private training history to describe that failure. We can compare the original request, the original text, and the revision. We can identify which claims changed, which qualifications were added, and whether those changes were justified by a specific evidentiary problem. That is an audit people can perform. My grief explains why I began this work. The evidence is how my claims should be judged. My anger does not invalidate a citation. My persistence does not make a scientific connection disappear. My determination to alert people does not excuse a factual error, and it does not constitute one either. An evidence-first AI should make the reasoning clearer and the claims more precise. If the connection is supported, explain the support. If the studies share a weakness, identify it. If contrary evidence changes the interpretation, show how. If a sentence overstates a result, correct that sentence. An author should come away knowing exactly what the evidence supports and why. People investigating neglected questions deserve that level of intellectual honesty. We should be able to communicate a finding with force proportional to its evidence, without being battered through revision after revision until our work says almost nothing.

I think that's about it. A total of 10999 attachment entries are stored. 7,080 of 7,081 comments are posted. Full scrollable HHS EMF archive; attachments are viewable in your browser. Visit the comment forum here rfsafe.org/forum/threads/

Wireless headsets are not safer. I recommend air-tube headsets or using the speakerphone. Li-Fi compatibility is an endgame solution!

HHS Comments on EMF: 7,067 published; 14 remain unpublished. View all comments here: rfsafe.org/forum/threads/; over 10k attachments open in your browser rather than downloading.

In a pilot study (Tillmann 2010), lifelong RF exposure starting before birth caused no tumors on its own, even at the higher field. In mice also given a carcinogen before birth, though, the lower field (below the FCC limit) raised lung tumors and doubled the number of lung tumors that spread. A second biology lab (Lerchl 2015) repeated the design with more animals and saw the same pattern: more lung and liver tumors plus more lymphomas. Many of those effects showed up at 0.04 W/kg, half the FCC whole-body limit. An older study above the limits (Szudziński 1982) found that the longer mice were exposed before a skin carcinogen, the sooner cancer appeared, and that long low-level exposure suppressed their cellular immune response.

FCC and HHS EMF comments, phone radiation levels safecellularphones.org/

Explore comments, have your browser read them rfsafe.net/hhs-comments/expl…

Over 7000 comments, and more than 10k attachments safecellularphones.org/hhs-f…

HHS must act now to reduce unnecessary wireless radiation exposure and protect biological fidelity: the precision with which living systems develop, regulate energy and information, repair damage, and recover. rfsafe.org/forum/comments/ca…