Luis Leonel Leon
It does not take an outright lie to distort the meaning of a news story. Sometimes it is enough to decide which truth belongs in the headline, which one is pushed into the summary or buried in the body, and when attribution quietly disappears. That is what happened in part of the coverage of the leaked Hope Florida grand jury report.
The document is severe. It says the grand jury concluded that $10 million from Florida’s settlement with Centene had been “misappropriated” and used for political purposes. To omit that would distort what the report says. But so would ignoring how the report ends: “we find insufficient evidence to charge anyone criminally.”
The grand jury maintains that misappropriation occurred, yet also acknowledges that it could not prove who was responsible or gather sufficient evidence to charge anyone criminally. Reading that, I found it difficult not to ask a basic question: How can such a serious conclusion be sustained while the same report acknowledges that there is not enough evidence to charge anyone or establish who was responsible? That legal tension, which was not raised in the coverage examined here and which I have not seen addressed elsewhere, deserves separate analysis by legal specialists.
A little background is useful. The investigation began with allegations made during Florida’s 2025 legislative session by Alex Andrade, the Republican representative for House District 2, who went so far as to allege possible criminal conduct. The matter was later referred to the State Attorney’s Office for the Second Judicial Circuit, headed by Jack Campbell, a state attorney elected as a Democrat. The Florida Bar distinguishes between a grand jury report or presentment, which may contain findings and recommendations, and an indictment, which initiates a criminal prosecution. Florida law expressly contemplates reports concerning individuals that are not accompanied by an indictment. The Hope Florida document may therefore carry institutional authority, but it is not a judgment and does not amount to a conviction. Yet in several media outlets, a report that produced no charges and contained insufficient evidence to charge anyone criminally became something closer to a media verdict.
After the leak became public, Gov. Ron DeSantis rejected the grand jury’s characterization. He said there had been “no diversion of any Medicaid funds,” defended the AHCA settlement as legally sound and in the state’s interest, and described the $10 million as a private contribution made within a private agreement. But the issue goes well beyond DeSantis and the $10 million. In my view, the episode exposes a deeper problem in contemporary journalism: the erosion of the discipline that should separate what happened from what was alleged, what was alleged from what was concluded, and what was concluded from what was proven.
Journalism already has simple tools for preserving those distinctions: “according to,” “states,” “alleges,” “concludes,” “could not be established,” “no charges were filed.” These are not ornaments. They tell readers the actual degree of certainty attached to a claim.
Just days earlier, I had read From “May” to “Is”: Certainty Distortion in Language Model Rewriting, a recent study by Catarina G. Belem, Shang Wu, Hongyu Yao, Mark Steyvers, Sameer Singh, and Padhraic Smyth, published as a preprint on arXiv in June 2026. The authors describe “certainty distortion” as a change in how confidently a statement is expressed while its essential meaning remains largely intact. When I encountered these stories, the connection was hard to miss. In the cases examined here, the shift runs in one direction: toward exaggerated certainty.
First we read, “according to the grand jury, there was a diversion.” Then, “the grand jury concluded there was a diversion.” Finally, we arrive at “the $10 million diversion.” Nothing has changed except the disappearance of the source.
CBS News Miami, which obtained and published a copy of the report, offers a clear example. Its headline read, “Grand jury finds DeSantis administration ‘misappropriated’ $10 million in taxpayer money to Hope Florida.” CBS told readers early in the story that there was insufficient evidence to charge anyone criminally. Later, however, the article referred to “the diversion of $10 million in taxpayer money” and used language that could be read as “diverting” or “siphoning off” funds to describe the alleged scheme.
A semantic shift occurs. “The grand jury says there was a diversion” becomes simply “the diversion,” even though no new evidence has appeared between one formulation and the next. Attribution disappears. The conclusion remains, now with the appearance of established fact.
The Guardian used a similar construction: “Florida grand jury finds DeSantis administration ‘misappropriated’ funds from Medicaid settlement,” although its treatment was more careful. The summary notes that the grand jury declined to indict anyone, and the opening paragraphs present both the finding of “misappropriation” and the insufficiency of evidence to bring charges. There is no concealment. There is hierarchy.
That distinction matters. A 2025 study published in Nature Human Behaviour, based on more than 35 million Facebook posts, found that roughly 75 percent of shared links were circulated without the person first clicking on them. That does not mean exactly 75 percent of readers consume headlines only. But it does show how easily a news story can circulate without the underlying article ever being opened. The person who sees only the headline receives the accusation first. The person who opens the story may later encounter its principal limitation.
That raises another uncomfortable question: How much of this can happen without premeditation, simply through habits and formulas journalism has absorbed over time?
Part of it may reflect the ideological orientation of a journalist or the editorial line of a publication, which are not always the same. A publication may define itself as conservative or centrist and still run a piece whose political framing points in another direction. A journalist may work within an editorial line different from his own. Or the problem may arise from a professional grammar reproduced through routine, speed, imitation, and the economics of headline writing.
The absence of intent, however, does not erase the effect. If a substantial share of news circulates without readers opening the article, what a newsroom places in the headline acquires disproportionate informational weight.
There is another dimension, perhaps an even deeper one: the effect of a headline does not necessarily disappear once the reader opens the story. Ullrich Ecker, Stephan Lewandowsky, Ee Pin Chang, and Rekha Pillai demonstrated this experimentally in a 2014 study published in Journal of Experimental Psychology: Applied. They found that a headline can induce a misleading interpretation without containing an outright falsehood, and that headlines shape how readers process the information that follows, affecting memory, inference, and even behavioral intentions.
Perhaps more important for this discussion, later information does not always succeed in correcting the impression created by the headline. A headline is not merely the doorway into a story. It can become the cognitive frame through which everything that follows is interpreted.
NBC 6 South Florida and WUSF used almost identical grammar. NBC headlined, “Leaked Hope Florida grand jury report says state ‘misappropriated’ funds,” and placed immediately beneath it: “The report found insufficient evidence to file criminal charges against anyone.” WUSF used, “Leaked Hope Florida grand jury report: State ‘misappropriated’ funds,” while also noting in the summary that there was insufficient evidence for charges.
These stories are more balanced than their headlines might suggest when read in isolation. But they still reveal an epistemic asymmetry: the most visible statement acquires greater certainty and relevance than the qualification that limits it. In digital journalism, where a headline can circulate independently of the article, unequal hierarchy is no longer merely a stylistic choice.
Associated Press headlined, “Florida grand jury finds DeSantis administration misused $10 million, but no charges filed.” Even there, the structure presents misuse as the headline’s principal certainty and the absence of charges as the contrast, while the grand jury itself said it could not determine who was responsible.
FOX 35 wrote, “Grand jury: $10 million from Hope Florida was ‘misappropriated,’ but no one was charged.” Florida Trident wrote, “Sealed for seven months: grand jury found $10 million ‘misappropriated’ for Hope Florida, but no one will be charged.” In both cases, the absence of charges remains visible, but the tension between the finding and the inability to attribute individual responsibility is compressed into a brief formula.
Florida Trident goes further when it later writes that the panel had “concluded that the money was taken,” a more categorical formulation than simply attributing “misappropriation” to the grand jury. The grammar makes misappropriation the dominant certainty and the absence of an identified responsible party the qualification. In my view, that hierarchy can lead readers to perceive as established a personal attribution the grand jury itself did not prove.
Action News Jax presents a different problem. “Grand jury says DeSantis administration ‘misappropriated’ $10 million in taxpayer money for Hope Florida” preserves attribution. But in the body, the story states that the money was supposed to provide health insurance for poor children and only afterward adds “the report says.” The source is present, but the sentence structure turns a disputed legal interpretation into the main proposition.
Andrew Sheeran, then general counsel for the Agency for Health Care Administration, testified that Centene had identified $56,248,611 in potential Medicaid-related damages and approximately $10.8 million in additional possible damages and settlement incentives. Under his interpretation, the $10 million directed to Hope Florida did not constitute recovery of Medicaid funds. He also defended that reading on the grounds that the settlement occurred before a lawsuit and that payment was made directly to Hope Florida.
The grand jury rejected that interpretation and concluded that the entire settlement represented reimbursement to taxpayers. I do not intend to decide here which legal interpretation is correct. The relevant point is that the dispute existed within the record itself. To present one interpretation as an undisputed premise is to resolve the controversy before explaining it.
The Miami Herald introduces another mechanism. It headlined, “Hope Florida grand jury findings revealed despite efforts to keep report private,” and immediately noted that the grand jury had identified Attorney General James Uthmeier as the “nexus” of much of the activity surrounding the transfer.
Neither statement needs to be false for the two, placed together, to create a more serious impression than either would produce alone. The reader first encounters the idea of a report someone tried to keep out of public view and then, immediately afterward, the name of an official placed at the center of the operation. The effect does not depend on expressly accusing Uthmeier of a crime. It is enough to arrange the facts in a way that suggests personal responsibility the criminal process did not establish.
Here the exaggeration of certainty does not arise from disappearing attribution. It arises from narrative proximity. Two true statements, placed side by side, can induce a third conclusion that neither establishes on its own.
Florida Politics, by contrast, chose a different frame: “DeSantis’ office lashes out over alleged leak of Hope Florida grand jury report.” The article reports early that the grand jury characterized the funds as “misappropriated,” but immediately adds that it recommended no charges and that no one assumed responsibility for directing the money toward Hope Florida.
Its treatment shows that different newsrooms can work from the same document and preserve, to different degrees, the distance between an institutional conclusion and established individual responsibility. There was no single inevitable way to tell this story. The result depended in part on editorial choices, professional habits, and the degree of caution each outlet brought to the task.
From these ten pieces, I draw a conclusion more precise than the claim, simplistic to some and obvious to others, that “the media lied.” There is no evidence of coordination among newsrooms, and they did not all behave alike. What can be documented is that several of the stories distorted the degree of certainty attached to the underlying facts.
They began with a grand jury conclusion and, through disappearing attribution, more categorical formulations, unequal hierarchy, and narrative association, presented it in ways that could make individual responsibility appear established when the grand jury itself had failed to prove it.
That, in my view, was the problem.
Reporting that the grand jury concluded that misappropriation occurred was correct. Turning a limited institutional conclusion into a more definitive narrative than the absence of identified responsible parties and the lack of sufficient evidence for criminal charges allowed is something else.
This pattern has a familiar theoretical framework in communication studies. In his 1993 Journal of Communication article “Framing: Toward Clarification of a Fractured Paradigm,” Robert M. Entman explained that framing selects certain aspects of reality and makes them more salient in order to promote a particular problem definition, causal interpretation, or evaluation.
Framing need not rely on falsehoods. It can operate entirely through true facts. The decisive choices are what receives emphasis, in what order, and with what weight.
Repetition adds another dimension. In a 2015 meta-analysis published in Journal of Advertising, Susanne Schmidt and Martin Eisend found that repeated exposure increases message recall. Their subject was advertising, not political journalism, so the findings should not be imported mechanically. Even so, they support a relevant point: what is repeated has more opportunities to be remembered.
That matters when “misappropriation,” “diverted Medicaid funds,” “DeSantis administration,” Hope Florida, and Casey DeSantis repeatedly occupy the most visible spaces while “insufficient evidence to charge anyone criminally” receives less prominence.
Many of the headlines use a seemingly innocent word: “finds” or “concludes,” as in “the grand jury concludes.” That is legitimate language, but it is not equivalent to “proved in court.”
The distinction matters because the grand jury itself acknowledged that it had not gathered sufficient evidence to charge anyone criminally and could not prove who was responsible. It is not the same to write, “the grand jury concluded that misappropriation occurred,” as to state, “the DeSantis administration diverted $10 million.”
The first preserves attribution. The second turns the conclusion into an autonomous narrative fact. The grammatical difference may be slight. The difference in certainty is not.
CBS headlined, “Grand jury finds DeSantis administration ‘misappropriated’ $10 million in taxpayer money to Hope Florida.” Associated Press wrote that the “DeSantis administration misused $10 million” and placed the absence of charges after the “but.” In both cases, the finding occupies the position of greatest certainty, even though AP makes the absence of charges visible in the headline.
There was no need to absolve DeSantis or discredit the grand jury. It was enough to place beside that accusatory conclusion what the same document acknowledges: there was insufficient evidence to charge anyone, and the grand jury could not prove who was responsible.
To avoid becoming captive to that exaggeration of certainty, I propose what I call the principle of proportional certainty: the degree of certainty expressed in journalistic language should correspond to the degree of certainty warranted by the evidence and the source.
If an institution alleges something, it should remain an allegation. If it reaches a conclusion, it should remain a conclusion. If there are indications, they should remain indications. When something has been proven, only then should it be presented as fact.
Journalism should not increase certainty beyond what the evidence warrants. Nor should it diminish certainty when the evidence supports a stronger claim.
In journalism, removing who is making a claim can also distort the nature of what is being claimed. If we write “according to the grand jury” because only the grand jury has reached a particular conclusion, the “according to” should not disappear without independent corroboration.
The principle of proportional certainty requires three things: attribution, distinction, and hierarchy.
Attribution keeps visible who is making a contested claim. Distinction separates an institutional conclusion from a criminal charge and a conviction, as well as evidence from inference and proof. Hierarchy prevents one of two relevant realities from being subordinated to the other when the same document both makes a severe finding and acknowledges that it cannot sustain charges or establish who is responsible.
Losing those distinctions has consequences. When an institutional conclusion is presented with greater certainty than the record allows, readers may come to perceive as established responsibility what was originally only an attributed and limited conclusion.
Public perception can shift. Political narratives can harden. Reputational damage can follow. Specific people can become associated with responsibilities that were never established. And a later clarification may arrive with far less force than the original impression.
The journalistic risk is to turn a difference in degree of certainty into a difference in reality.
In an era of social media, Shorts, and Reels, the standard should be more demanding because the headline may be the only part of a story a reader ever sees. A clarification several paragraphs later does not necessarily undo the impression created in a search engine, a notification, or a social feed.
For digital journalism, one simple rule suggests itself: a headline should be defensible even if it were the only part of the story a reader ever encountered.
This is not a rule designed to protect Ron DeSantis. Tomorrow it may protect a Democrat, a businessman, an activist, a public official, a journalist, or an ordinary citizen. That is precisely why it should be a rule rather than a partisan convenience.
In electoral terms, the effect is not neutral either. When coverage increases the perception of guilt or responsibility surrounding a Republican administration, the political cost falls primarily on its officials, candidates, and the party brand itself.
In the Hope Florida case, my reading is that the likely direction of that effect benefits Democrats and harms Republicans. That does not tell us how many votes might move, nor can its magnitude be measured without specific public-opinion research. But it does allow us to identify, reasonably, who bears the political cost and who may benefit from it.
If new documents, testimony, or evidence capable of identifying responsible individuals and sustaining criminal charges were to emerge tomorrow, the situation would change and journalistic language should change with it. If a new investigation produced an indictment or, later, a conviction, headlines could reflect those developments with all the force the facts warranted.
Until then, the obligation is to describe exactly what we know today: the grand jury concluded that misappropriation occurred, it could not prove who was responsible, and it did not gather sufficient evidence to charge anyone criminally.
Journalism cannot control what a grand jury concludes, what a politician alleges, or what a source leaks. It can control the degree of certainty with which it tells the public.
An accusation, conclusion, or inference begins to deform when it is written in the grammar of an accomplished fact. A headline should not become a public verdict no court ever delivered.
That, in some coverage, a report containing insufficient evidence to charge anyone and producing no criminal charges was transformed into a media verdict, with real political and reputational consequences, cannot be attributed to the document alone. Responsibility for that distortion of certainty, in my analysis, rests with the journalists and media outlets that chose to tell the story that way.
