A field of Roundup Ready soybeans can survive a glyphosate spray that kills susceptible weeds around the crop. The soybean has been changed to tolerate the treatment. Investigating the spray's effects on a worker requires a different inquiry, as does judging the contract for next year's seed.
Monsanto sold products that brought these questions together. Its seeds could make its herbicide more useful; the herbicide helped make the seeds attractive. A farmer could value the combination while objecting to its price or planting restrictions. A person could accept the food-safety findings and remain concerned about exposure to the spray.
Much of the argument becomes difficult to follow because GMO, Monsanto and Roundup are treated as names for the same thing. GMO describes an organism developed through genetic engineering; Monsanto is a company name and Roundup a brand used for formulated products. An answer about one can sound like an answer about all three, while leaving the important work undone.
A herbicide-tolerant crop survives an application intended to control weeds. It gives the farmer another way to manage weeds, without guaranteeing a larger harvest. The chemical can now be used where it would otherwise damage the crop.
Insect resistance works differently. Bt corn, named for Bacillus thuringiensis, produces proteins toxic to certain insect pests. The intended benefit is protection against those insects, which can reduce the need for sprayed insecticides. Herbicide tolerance and insect resistance can occur in the same crop, but one does not imply the other.
A single total for pesticide use can hide that change. A reduction in one insecticide and an increase in a herbicide concern different substances, targets and exposures. Their weight in kilograms alone cannot tell us the resulting health or ecological effect. The comparison needs to identify what changed and what the farmer would otherwise have used.
The FDA's account of why farmers use engineered crops describes weed control with less tillage, as well as protection against insects and plant viruses. Less soil disturbance can reduce erosion. Whether a particular farm benefits depends on its conditions and practices; the trait's name is not a guarantee of success.
Repeated reliance on the same control also creates a problem. The National Academies' 2016 assessment described weeds resistant to glyphosate and insects resistant to some engineered protections. The report tied such problems to management practices. The conditions that made a trait useful at introduction could change over subsequent seasons.
Consider what happens if a few weeds in a field can already survive the treatment. Spraying removes susceptible competitors and leaves the survivors a chance to set seed. Repeated use can favor their descendants. Selection under repeated spraying can explain that change without assuming a crop gene has escaped into every resistant weed. The resulting choices about other controls and their costs belong in an assessment of the system, alongside its initial convenience.
The loss of a useful weed-control method is a farming problem in its own right. Its consequences should be examined without first recasting it as a claim about poisoned food.
The same Academies assessment found no substantiated evidence of a difference in human-health risks between the commercially available engineered crops it reviewed and conventional crops. Its review considered food composition, animal studies and available human-health information. It acknowledged the difficulty of detecting subtle or long-term effects and did not claim that long-term epidemiological studies had directly measured engineered-food consumption.
The review gave no basis for treating the foods it assessed as more dangerous merely because they were engineered. It does not certify every future crop, every possible change in composition or every chemical used during cultivation. The committee recommended examining a plant's characteristics and potential hazards, including changes made through conventional breeding.
Consider a crop changed to alter its nutritional content and one changed to survive a herbicide. The relevant assessment of the food may concern different substances in each plant. The second also raises questions about the accompanying spray. Those plants need assessments suited to what has changed in each.
Food safety also does not answer who benefits economically. A crop can be suitable to eat while being sold under terms a farmer considers unfair. A criticism of those terms does not need a claim that the grain causes cancer in order to deserve a hearing.
Glyphosate is an active ingredient; formulated herbicides contain other ingredients too. Research can concern the active ingredient, a particular mixture or people exposed to products in working conditions. A result about one mixture or use needs to be read with that setting in view, rather than attached to every product carrying the brand.
In 2015, IARC classified glyphosate as probably carcinogenic to humans, Group 2A. Its explanation describes limited evidence in humans, sufficient evidence in experimental animals and strong evidence of genotoxicity, or damage to genetic material. The classification concerns the strength of evidence that an agent can cause cancer. It does not give a person's probability of developing cancer after a particular exposure.
Risk also depends on the amount, route and duration of exposure. This matters when moving between a person mixing and spraying a product and a consumer eating food with a residue. Calling both exposed says too little. Their situations cannot be equated without examining what reaches the body and under what conditions.
The regulatory disagreement cannot, however, be explained entirely by saying that IARC studies hazard while everyone else studies risk. EFSA's 2023 account of its assessment says it used the European Chemicals Agency's hazard classification, which did not classify glyphosate as carcinogenic. There was disagreement about the evaluation of carcinogenic evidence as well as about how exposure should inform regulation.
EFSA reported no critical areas of concern, a term it defined as a concern affecting all proposed uses and preventing approval or renewal. It also listed unfinished assessments and data gaps, including aspects of dietary risk and a component of the representative formulation. Its definition of critical concern allowed for those unfinished questions; the favorable headline did not mean that every proposed use had been cleared of every concern.
IARC says it evaluates evidence available for public scientific review, including industry studies that meet that condition. It does not exclude a study simply because a company paid for it. Access to enough information for independent assessment matters. Different selections and interpretations of studies therefore need to be examined alongside the different questions agencies ask.
The EPA's November 2026 account describes its earlier conclusion that glyphosate was not likely carcinogenic. It also records the 2022 court decision vacating the human-health portion of its interim review decision and the subsequent withdrawal of that decision. A new human-health assessment was still being prepared, with completion anticipated in late 2026.
The Agricultural Health Study offers one way to see what an investigation can contribute. Researchers collected information about pesticide use from licensed applicators and linked participants to cancer registries. People entered through an applicator cohort, rather than being selected only because they already had cancer.
The 2018 glyphosate analysis included 54,251 applicators in Iowa and North Carolina. Use estimates came from reported days and information about the intensity of application. The researchers accounted for other measured factors, including other pesticide use. They found no apparent overall association with solid tumors or lymphoid cancers, including non-Hodgkin lymphoma. They also identified evidence concerning acute myeloid leukemia in the most highly exposed groups that they said required confirmation.
The study gave different answers for different cancer outcomes. Exposure estimates can be imperfect, follow-up has an end date, and different diseases may give different results. Each result needs to be read with its estimate and uncertainty, including results inconvenient to the reader's preferred conclusion.
The dates also change the comparison. IARC's 2015 evaluation preceded this updated analysis. One cannot accuse the earlier panel of ignoring a later publication, or treat an earlier assessment as if no further evidence could affect it. The useful disagreement concerns which findings change the judgment and why.
A company can fund research that produces reliable results. It can also make its own contribution harder to see, leaving a reader to mistake a manufacturer's argument for an outside assessment. These possibilities need different responses: funding disclosure in the first instance, investigation of the concealed contribution in the second.
One prominent glyphosate review, by Gary Williams, Robert Kroes and Ian Munro, appeared in 2000. The journal later published a retraction notice, appearing in its February 2026 issue. The handling editor cited concerns about authorship, undisclosed contributions and possible compensation, as well as the review's selection of carcinogenicity studies. The notice says the surviving author did not respond to the concerns. It also expressly says the retraction does not itself take a position on whether glyphosate or Roundup causes cancer.
Concealed authorship damages the value of a review even before its individual conclusions are checked. An assessment offered as independent helps a reader judge a company's claims. If the company's role in writing it is obscured, the reader has been deprived of information needed to judge the assessment itself. A paper may repeat an accurate finding while misleading readers about whose judgment they are hearing.
Suppose an assessment cites the review for its account of an animal experiment. A reader needs to trace that account back to the experiment: the dose, comparison animals, recorded findings and method of interpretation. The review might have been a convenient summary, or it might have been the reader's only access to the result. If the underlying experiment is unavailable, the reader cannot simply repeat the check.
Conclusions that relied on the review need to be traced through the evidence supporting them. Removing one paper does not mechanically reverse every conclusion that cited it. Leaving it in place as ordinary, unqualified support after retraction would also be indefensible. EPA addresses this particular review on its glyphosate page. The agency says it identified the paper in its literature review but bases risk assessments on individual studies rather than directly on review articles. It says the retraction does not change its assessment. That is the agency's account of its reliance, not an independent check here of every underlying study.
The connection to Big Pharma and the Vaccine Conspiracy lies in this presentation of independence. Readers need to know who contributed to the judgment they are being asked to accept.
The ownership dispute concerns a different kind of control. In Bowman v. Monsanto, decided in 2013, the US Supreme Court considered a farmer who bought commodity soybeans from a grain elevator, anticipating that many would carry Monsanto's patented Roundup Ready trait. He planted them, applied glyphosate and saved seed from the surviving crop for later plantings.
Bowman argued that an authorized sale had exhausted Monsanto's patent rights. The Court distinguished using or reselling an item already sold from making further copies of the patented invention. It held that exhaustion did not protect his deliberate reproduction through planting and harvesting. The opinion limited its holding to the case before it, leaving other situations involving reproduction outside a purchaser's control unresolved.
This was not a case establishing that an unwitting farmer becomes liable whenever pollen arrives from a neighbor. Nor was the seed sterile. The ability to reproduce it was precisely why the dispute arose: biological reproduction and legal permission to reproduce had come apart.
The farmer's freedom to replant is still at issue. A farmer may be allowed to eat or sell a harvest while being forbidden to keep it for another planting. The patent holder regards the restriction as protection for an invention; the farmer may regard repeated purchases as a limit on autonomy and bargaining power. Deciding whether that arrangement is desirable requires attention to access, competition and the terms of sale. Toxicology cannot resolve it.
Dewayne Johnson's case concerned a worker's exposure to herbicide products and Monsanto's obligations toward him. The California appellate opinion describes a school grounds manager who developed non-Hodgkin lymphoma and prevailed at trial on claims including failure to warn. The jury awarded roughly $289 million in 2018. After reductions at trial and on appeal, the 2020 decision reduced the combined award to about $20.5 million while sustaining liability and entitlement to punitive damages.
The appellate court examined whether the record supported the judgment. Among the evidence it discussed were hidden company contributions to publications and Monsanto's response to questions about genotoxicity. It explained why the jury could infer disregard for safety; its role was not to conduct a fresh population study or calculate the risk from every use of glyphosate.
The judgment establishes legal accountability in that case. It is more than a public-relations setback, and less than a finding that every engineered crop is dangerous. Johnson's claim concerned what a worker should have been told about the product he used.
Bayer completed its acquisition of Monsanto in June 2018. Changing ownership did not make disputes over the products disappear. It also did not make the legal, agricultural and scientific questions interchangeable. A consumer considering a food, a worker applying a spray and a farmer signing a seed agreement need different information.
They share a reasonable demand: an account of the product that does not hide its conditions. That means identifying the crop trait, the exposure being assessed, the contributions behind a safety review and the rights attached to the seed. The company should have to answer those questions even where the evidence favors its product. A critic should be able to ask them without first claiming that the entire technology is a fraud.