51³Ô¹Ï

Subramaniam Vincent is director of journalism and media ethics at the Markkula Center for Applied Ethics, Santa Clara University. Views are his own.

 

Prediction Markets (PMs)–Kalshi, Polymarket, and more–are the latest ethical and regulatory mess to erupt in the digital media economy. But what should the media do about PMs? The questions are many. What does it mean when news anchors and articles cite the odds numbers for various future events? What should the media call these odds? Are they really probabilities or likelihoods or something else? Should media organizations take money from PMs in return for offering ticker space to showcase the latest “odds”? Should reporters and editors just turn the whole PM space into a beat like some news outlets (e.g. Wired) are?

PMs are both the topic of news cycles and also colliding with the news work itself. Multiple dilemmas are emerging. But before we name the questions, and develop an approach to address them, a note on the context, since media policy cannot be developed in a vacuum. We need to start with the legal and regulatory situation.

While this article is somewhat U.S.-centric, it is best to begin with what’s happening in Europe, where there is less of a brouhaha. Most European nations appear to have a clear-eyed consensus that PMs are indeed gambling. Once you define PMs as gambling, pre-existing regulatory stances and actual regulation are able to drive policy decisions and rule development. Many regulators or governments have simply outright banned Kalshi and Polymarket or geoblocked users from their countries.

In the U.S., the situation is wildly different. First, there is a definitional dispute sprawling out across federal-state power line. Lawsuits are being filed left, right, and center. Over a dozen U.S. states have sued, primarily on grounds that the activity on PMs is gambling. Arizona even filed a criminal case against Kalshi. The federal futures regulator CFTC has stepped in on behalf of Kalshi and Polymarket trying to preempt the jurisdiction of the states and threatening to sue state regulators. (On the regulatory stance from the U.S. United States Commodity Futures Trading Commission (CFTC), one note: CFTC is supposed to have five commissioners, but it is led by one Trump appointee who is the chairman, and four seats are currently vacant. U.S. president’s son Donald Trump Jr. is connected with both Kalshi and Polymarket, and there are reports the Trump organization plans to launch its own PM.)

Second, prediction markets are also in the news because major episodes have emerged on either suspicion of or outright prosecution for insider trading (Maduro’s capture, Iran war, etc.), and betting actors trying to influence outcomes or pressure journalists to confirm or disconfirm events one way or another.

It’s worth noting the main attraction for PMs are that they are secondary markets, designed to invite everyday people to take and trade bets on future political events. Such events are often ones where capable journalists and analysts may report on with deep sourcing, so the risks of manipulators or opportunists stepping in is real. And given the temptation of journalists themselves (to place bets) when they have private knowledge of political moves before their stories go public, it isn’t surprising that ProPublica took the lead recently and barred its journalists from participating in PMs.

Third, PMs have also become a revenue opportunity for news organizations as “data sources” for “probabilities” of future events, particularly for election outcomes, political candidate prospects, and so forth. For broadcast and big new media organizations, polls and pollsters have been a part of news investments. Horse-race stories–the routine reduction of election coverage to poll standings–have long been criticized by ethicists, and yet they remain a staple of the news cycle. PMs have now been added into the mix because of their “forecasting value.” A whole range of mainstream news orgs have tied up with both Kalshi and Polymarket to either list their betting odds as data tickers on the same screens where they show poll results, or use their data in other ways. The PMs are paying the media companies in what can only be called “paid sourcing” or “brand placement”–all of which is a way for the PM players to legitimize their products through the media and expand their betting user bases.

But the news media is not merely implicated in ethical issues on their evolving relationship with the PM platforms as data sources or temptations on journalists to monetize their private knowledge. There are multiple and deeper reasons for media company executives and journalists alike to parse the public worth of the claims PM promoters make. One example emerges from something both news reporters and PMs share: the capacity to reduce information asymmetry.

When reporters pull in privately or semi-privately held knowledge and information from credible sources and make them public through stories, they are reducing information asymmetries to that extent. Stories carry their facts, claims, and beliefs of their sources and are generally accessible to everyone in discourse (setting aside paywalls for the sake of the principle that news is supposed to be public good.) PMs are incentivizing bettors with particular knowledge of the likelihood of a future event to weigh in with their money. These could be complex and consequential future real-world events that can impact the lives of thousands to millions of people. This likelihood numeric is the price and is posted on the platform for everyone to see. Therein lies the PM claim to bring in privately-held beliefs about the likelihood of future events.

These live odds factor in not just such knowledges, but also the sentiments and beliefs of people who may not have specific knowledge. These people may simply be betting on their beliefs drawn from a narrative they subscribe to: intuition, expert sense, and so forth. Entire swathes of bettors on Kalshi and Polymarket are there for the ride (secondary markets) like short-term day traders, and hence there is credence to the claim by regulators around the world that this is gambling.

For this and more reasons as the article will show, PMs are going to be a deeper challenge for the news media. There is a serious imperative for news media to run 101-type explainers on what PMs are to the public. One way to do this is to compare PMs definitively with the polling industry at the level of a primer. People are familiar with polls. PMs are new and their odds are being showcased in broad news segments. So a comparative approach may help.

The media ethics program did this comparison recently to build a primer and a decision-making review matrix for media organizations on prediction markets. The second part will tabulate some decision-making challenges in newsrooms relating to PMs. For each of these challenges, we’ll look at what is at stake and the respective ethical issues more granularly. The hope is that this can then be used as a starting point for review in news organizations on policy to deal with the challenges.

Before that a note of history. PMs are not new. They have long been a research topic for information and market theorists, people studying information asymmetries, epistemics and the role markets can play in mitigating them. The University of Iowa has been operating a PM called the Iowa Electronic Market (IEM) for over two decades now, specifically on political bets, election outcomes and so forth. While the goal has been both academic research (many universities are involved) and student education, the fundamental position limit of $500 for bets is a structural safeguard that is less well known. IEM also runs as a secondary market with traders’ bids and asks being matched by the platform. The research that emerged on the accuracy of PMs for some national election forecasting use cases (where PMs have outperformed polls) has made IEM the academic and real model, as well as forerunner to Kalshi and Polymarket.

This broad sphere of studies and findings about PMs stretching back over two decades could be marshalled for norms and policy development in media organizations.

Prediction Markets vs. Polls

Polls and prediction markets are not the same, and currently in the rash of news reporting on the disputes and manipulation cases, the news media has a great opportunity to explain PMs to the public in comparison with polls. Do polls and PMs measure the same thing? What’s different? How accurate are polls vs PMs? How do the structural weaknesses and ethical pressures compare? What are the risk factors for, in and of news media itself?

Below, we work through the key differences by doing a category-by-category comparison. It helps view what PMs are doing in much deeper, structural light than the glossy claims they make about themselves as “truth machines” or “accurate forecasters.”

What is being measured?

Polls measure the stated preferences or opinions of a sampled population. When a polling firm asks, “Who do you intend to vote for?” it is trying to capture what people say they plan to do, on the assumption that stated intention predicts behavior. The formal claim about a poll’s informational value runs through an inference chain: stated preference → probable behavior → aggregate outcome.¹ However, polls have well-documented vulnerabilities, especially when what people say they prefer does match their actual behavior when the times comes. (More on this later.)

Prediction markets, by contrast, do not measure opinions or voting intentions. They measure the financially weighted probability beliefs of bettors. When a contract on Kalshi or Polymarket trades at $0.60, this means the market’s consensus is that there is approximately a 60% chance the specified event will occur. This is a probability belief, yes, but this number is not a simple average of what individual participants (bettors) believe. Wealthier traders can express stronger conviction by committing more capital. So, the price technically reflects what economists call a wealth-weighted average belief—the precise term used by Wolfers and Zitzewitz in their foundational analysis—rather than an equally-weighted mean.⁷

Furthermore, a lot can happen between the opening of an event market and its actual resolution, that involves decisions by people, leaders, organizations, unexpected events, and so forth. PMs are secondary markets that then “track” what is going on through the belief lens of the betting crowds. So, PM estimates of the likelihood of some event happening (resolving to “true” in PM parlance) is not a straightforward probability estimate.⁸

Notes for news reporters:

The problem is also not simply that polls get it wrong too. News media have been criticized for horse-race journalism’s effects on audiences, for instance over effects on turnout and vote choice. ² Also, research on the predictive validity of polls shows that their accuracy as forward-looking instruments is highly time-dependent. Polls conducted far from an election carry much less predictive information than those conducted in the final week; stated preferences at nine months out have low correlation with the final vote.³

Major comparative studies, analyzing over 30,000 national polls from 45 countries between 1942 and 2017, document systematic polling errors across time and context, especially in plurality electoral systems.⁴ In media coverage, criticism of “horse-race journalism” stems precisely from this gap: polling as a snapshot is often published as a prediction, misleading audiences about electoral certainty.⁵ Some research has also shown that probabilistic forecasting derived from polls can suppress voter turnout by overstating the certainty of an outcome.⁶

A PM contract price of $0.60 may look similar to a polling score like 60% and on the broadcast ticker of a TV screen, both expressed as a percentage. But they are measuring categorically different things. The poll reflects what a sampled cross-section of the relevant population says it will do. The PM price reflects what financially motivated bettors, weighted by their capital, believe will happen.

Treating them as interchangeable is a category error. This means the media is conflating meanings, impeding public understanding instead of fostering it.

What is the unit of aggregation (measurement)?

In polls, each responding person is one equal data point, but each person’s opinion does not necessarily get equal weight in the final result. The goal is proportionate representation, not strict equality.⁹ The gold standard in professional polling is probability-based sampling followed by post-stratification weighting, in which respondents are weighted up or down to align the sample with known population proportions on demographics such as age, race, gender, and education. When sample sizes are small, accuracy degrades and the margin of error widens in a statistically quantifiable way.¹⁰

In prediction markets, the bettor is the respondent, but each dollar has one unit of belief-weight. If you bet $10,000 instead of $10, you are committing more capital to a position and exert greater influence on the final price than smaller bettors holding the same belief in the likelihood of the event.

Bettors also have different risk profiles and tolerances. Charles Manski’s analysis showed that the price is a belief distribution among traders and its exact location depends on the distribution of wealth and risk preferences in the market. When risk preferences come in, everything changes. If you are very wealthy, a $10,000 bet maybe something you can lose and not break the bank, and so you can wager more money on bets for the odds you see posted. If I am not, even a $500 loss may make default on other crucial needs or obligations, so the risk tolerance also drives bet sizes. As Manski showed, this means a PM price does not simply aggregate “what people think”—it aggregates what people think, scaled by how much money they are willing to stake on it.⁸

The market is therefore capital-weighted, not person-weighted.

In thin or low-liquidity markets—such as local and regional event contracts now appearing on Kalshi—a single well-capitalized trader can move the price substantially. Given that, both journalists and everyday people need to know that those probability estimates are not population-level measures ¹¹ and should not be accorded that status.

How do claims of accuracy for polls and PMs compare?

Polls claim their value as representative snapshots of mass opinion at a given moment. But their record as predictors of election outcomes is mixed. The most consistent finding in the literature is that polling accuracy improves significantly the closer polls are conducted to election day. Early polls are poor predictors; late polls are better, but still subject to systematic error. The 2016 and 2020 U.S. presidential election failures—were driven by social desirability effects and non-response bias, where certain voter groups systematically declined to participate, masking actual voter intentions.¹²,¹³

Prediction markets have been reviewed by scholars, and especially on electoral margins forecasting, they have outperformed polls. The University of Iowa’s Iowa Electronic Markets (IEM)—the longest-running academic PM, operating since 1988, outperformed polls 74% of the time overall for 13 political candidacies from 1988 to 2004.¹⁴ (IEM does not run markets on U.S. electoral college outcomes in the presidential election, only on the popular vote and vote shares.) But the accuracy advantage for PMs over polls is narrowing. American Association for Public Opinion Research (AAPOR)’s post-election review of the 2024 U.S. election cycle found meaningful improvements in polling accuracy compared to 2020, reducing the gap ¹⁷.

One noteworthy point: IEM’s accuracy record is indeed an empirical basis for Kalshi and Polymarket’s accuracy claims, but it was built under strict structural constraints: a $500 maximum position per trader, academic governance, and restriction to electoral vote-share contracts. Kalshi and Polymarket, on the other hand, operate without these constraints, at vastly higher volumes (and dollar bets) across a far wider range of event types. Their accuracy on non-electoral topics remains largely untested in peer-reviewed literature.¹⁴

Note for news reporters

PMs do outperform deliberating groups and expert panels on forecasts. They eliminate the cascade effects and social pressure biases that systematically skew group judgments, because each trader acts on private information with money at stake rather than social approval.¹⁵

What is the incentive structure for “accuracy”?

Polls capture the stated preferences of people at a moment in time. Respondents face no financial penalty for inconsistency or misrepresenting their own views intentionally, or wishful thinking. Survey methodology research has long documented “satisficing”—a term coined by Herbert Simon and applied to survey behavior by Jon Krosnick—where respondents settle for a good-enough answer rather than reflecting carefully, particularly in long or complex surveys.¹⁹ There is no mechanism to reward a respondent who turns out to have been right, nor to penalize one who was wrong.

Prediction markets create strong incentives to bet on genuine private beliefs, seek out accurate information, and update positions quickly as new information arrives. The profit motive specifically rewards contrarian, privately-held knowledge: a bettor who knows something the market has not yet priced can profit by acting on it, which in turn moves the price toward a more accurate level. Theorists describe this as PMs’ capacity to surface and aggregate information asymmetries²¹—knowledge dispersed across many individuals that no single analyst or newsroom possesses.²⁰

This incentive argument is theoretical. It assumes that bettors are primarily motivated by accurate forecasting. In practice there are real differences depending on the specific bets, who is participating and why. The Eric Swalwell case in the California Governor’s race is a good example. Until the sexual misconduct allegations exploded on the scene, it appeared that bettors were going with a political narrative to wager their money on the likelihood of Swalwell winning.

A significant share of PM participants–particularly on Polymarket, which operates via cryptocurrency–appear motivated by speculation, entertainment, or ideological conviction rather than private information. When uninformed or sentiment-driven traders dominate a market, the price signal degrades regardless of the financial incentive structure.²²

It appears, fundamentally, what PMs are really doing is allowing bettors to monetize information asymmetry, and to legitimize the monetization, the products create a larger reservoir of lesser informed bettors to constitute the market. The ethics of this for public policy, elections, war, and governance-related bets is astoundingly problematic. Take the “wisdom of the crowds” framing being used to claim that PMs are legitimate. If the price of a critical referendum of ballot measure passing is $0.7, does the betting crowd that has created that price have some privately held and shared knowledge that the crowd that is betting the other way does not? Or vice versa? If this is a public ballot measure has consequences for people in everyday lives, why is that information not already public for everyone? But if the $0.7 crowd shares that with the $0.3 crowd, there is no asymmetry anymore and the bet will dissolve, and people would have lost money by then.

The idea that public policy decisions–where as people, we govern ourselves collectively through elected representatives–is being subject to hidden knowledge and asymmetric wagering is undemocratic and hence was always going to need legislative review.

Deeper: What are some structural weaknesses?

Design

Polls can ask open-ended, multi-part, or scaled questions. They can capture nuance, ambivalence, and degree. Prediction markets cannot. Every PM contract must be binary, precisely defined, and unambiguously resolvable: did the event happen or not, by a specific date, as defined by specific criteria? PMs are structurally unsuited to questions that involve abstract policy assessments, normative judgments, retrospective analysis, or contested factual disputes.

This is by design—it is the mechanism that makes financial settlement possible between winners and losers. The result? Complex, contested, or context-dependent realities get structurally flattened into yes/no propositions.²³

In 2003, the Pentagon shut down a terrorism futures experiment called Policy Analysis Market (PAM). Why? The goal was to convert complex geopolitical questions into binary contracts, and this required custom mathematical architectures just to define resolution criteria. Furthermore, it could not accommodate the conditional dependencies of real-world security events.²³ That led to its collapse. There were documented PM failures on individual decision-maker questions. For instance, markets wrongly found it more likely than not that Karl Rove would be indicted in 2005, and that John Roberts would not be appointed to the Supreme Court. This exposed the limit: when the relevant information is held by one or two individuals rather than dispersed across many, there is no crowd wisdom to aggregate.¹⁵

Liquidity and thin markets

Both polls and PMs degrade when their input samples or bettor base is too small. The difference lies in how they fail. In polling, a small sample produces a wider, statistically quantifiable margin of error—the problem is visible and disclosed because how polling companies have agreed to methodology and transparency standards.

In prediction markets, a thin market produces a price that may look precise but carries no reliable signal. Between 66 and 75 percent of new futures contracts ultimately fail due to insufficient trading activity.¹¹ In a low-liquidity market, the last recorded trade may be stale, and a single large trader can move the price without any new information entering the market.¹ A good example of thin markets on PMs currently are local bets.

PM promoters may counterclaim that their concept adoption is still at an early stage, and bettors for local predictions will come sooner or later, improving liquidity. But this could be more complicated than that. Would community bettors be as detached when wagering their money on something local that actually could impact their neighbors and themselves the way they are when placing bets on far-away events. These are open questions.

Note for news reporters:

Local events (and therefore implications for local newsrooms) are currently very low-liquidity. Kalshi has begun listing local and regional event contracts—municipal races, state ballot measures, local policy outcomes—where trading volumes are very low, often a few hundred dollars. It is not clear that such prices have any reflection of community-level information. Reporters in particular should be cautious about treating low-volume local PM odds as meaningful signals, let alone newsworthy data.¹

Can participants exploit private knowledge to gain unfairly or manipulate outcomes?

In polling, coordinated misrepresentation—where people deliberately state a false preference to skew published results—is theoretically possible but structurally difficult. It requires large-scale coordination, offers no direct financial reward to the individual respondent, and professional polling organizations are structurally independent of outcome actors. Strategic response to polls is a documented but limited pathology.²⁵

Prediction markets are creating a news cycle in 2026 all by themselves on exactly this. Anyone with advance knowledge of a future event—a government official, military officer, corporate executive, or journalist—can potentially place a bet and profit directly from that knowledge. This is not a hypothetical concern. It is now a documented and prosecuted reality.

On April 23, 2026, the U.S. Department of Justice unsealed a federal indictment charging Master Sergeant Gannon Ken Van Dyke, a U.S. Army Special Forces soldier involved in the military operation that captured Venezuelan President Nicolás Maduro. It has charged him with multiple types of fraud, theft of nonpublic government information, and unlawful use of confidential information for personal gain. Van Dyke allegedly made more than $409,000 using a Polymarket account on bets totaling $33,000, taking “Yes” positions on contracts including “Maduro out by January 31, 2026” and “U.S. Forces in Venezuela by January 31, 2026.”

An Israeli military reservist and a civilian were previously indicted for using classified information to bet on the Israel Iran war.²⁴ Senator Chris Murphy cited blockchain analytics identifying wallets that collectively earned approximately $1 million betting on U.S. military strikes on Iran shortly before the strikes occurred.²⁷ In each case the mechanism was the same: an actor with privileged foreknowledge used a PM platform—which carries no legal prohibition equivalent to SEC insider trading rules—to convert classified or confidential information into private profit. (See more on this in section below on regulation).

Regulation and definitional disputes

Polling in the U.S. is largely self-regulated. There is no federal licensing requirement to conduct polls. The governing standards are AAPOR’s voluntary code of professional ethics, FEC disclosure requirements for campaign-commissioned polls, and general consumer protection law. Polling firms are not regulated as financial entities. No state attorney general has sued a polling firm.⁹

On prediction markets in Europe, for the moment, the regulatory consensus is clear: prediction markets are gambling. France’s Autorité Nationale des Jeux (ANJ) concluded in 2024 that Polymarket constitutes unauthorised gambling, describing its products as having “addictive characteristics like those found in online gambling,” and geoblocked French users.³¹

Portugal’s gaming regulator (SRIJ) banned Polymarket in March 2026, stating it was “not authorised to offer betting in Portugal” and that “betting on political events is not permitted under national law.”

Hungary, Belgium, Germany, the Netherlands, Switzerland, Romania, and Poland have all blocked or blacklisted Polymarket as unlicensed gambling.³²

In the UK, the BBC cannot accept advertising from PM firms because UK regulators classify them as gambling products—a decision with direct implications for any public broadcaster considering PM data partnerships.³²

In the United States, whether PMs are gambling or something categorically new has become the defining legal question and dispute of 2025 51³Ô¹Ï–2026.

Nevada’s Carson City District Court banned Kalshi sports contracts in April 2026; the Nevada Gaming Control Board has stated that “prediction markets, to the extent they facilitate unlicensed gambling, are illegal in Nevada.” Arizona’s Attorney General Kris Mayes filed criminal charges against Kalshi for illegal gambling. As of early April 2026, 14 of 16 state preliminary rulings had gone against prediction markets, with only New Jersey and Tennessee siding with the platforms.³³

The federal picture is the reverse.

One challenge is jurisdiction. The Trump administration’s CFTC has asserted that prediction markets fall exclusively under federal jurisdiction as commodity futures contracts, and in April 2026 sued Arizona, Connecticut, and Illinois to block state-level enforcement. The CFTC chairman declared PMs under exclusive federal authority.³⁴

There is also a political dimension: Donald Trump Jr. is connected to both Kalshi and Polymarket, and reports indicate the Trump Organization is planning to launch its own prediction market platform. It would appear to the casual observer that the CFTC’s aggressive federal preemption stance reflects both regulatory philosophy and political alignment.

The second is a regulatory challenge: bettor identification. The class of people who can influence outcomes relevant to PM contracts—military personnel, intelligence officers, government officials, corporate executives, and journalists—is not a defined or registered “category”. Kalshi, as a CFTC-regulated entity, requires Know Your Customer (KYC) identification for all users. (CFTC rules have not historically applied SEC-style insider trading prohibitions to event contracts.) Polymarket is incorporated offshore and hence not CFTC-regulated, and operates predominantly via cryptocurrency wallets, with minimal KYC for non-U.S. users. Note that Van Dyke used a VPN to disguise his location. KYC rules will never be enough.

Under congressional pressure, Kalshi and Polymarket did announce voluntary guardrails in March 2026—banning political candidates from trading their own races, prohibiting trading on confidential information, and adding transaction surveillance tools. But category-based bans—such as prohibiting candidates from trading on their own races—are enforceable only where the platform can verify identity. The announcements from Kalshi and Polymarket came only after the introduction of the Death Bets Act and the Prediction Markets are Gambling Act, both filed in March 2026.²⁸ That said, for the wider class of outcome-influencers, while blockchain analysis and transaction pattern surveillance are usable for detection, both are reactive rather than preventive.^24,26

The Iowa Electronic Markets addressed this structurally from the outset: a $500 maximum position per trader, academic governance, and restriction to electoral vote-share contracts made manipulation both economically irrational and institutionally monitored. Wolfers and Zitzewitz note that despite documented attempts to manipulate IEM prices, none had a discernible effect on outcomes. Kalshi and Polymarket have deliberately discarded those safeguards in pursuit of scale. With the Van Dyke indictment, the CFTC filed its own parallel civil action—the first time the agency has pursued a PM-specific insider trading case.^{24 (}The CFTC’s new Director of Enforcement listed insider trading in prediction markets as a top enforcement priority ²⁷.) Meanwhile, the Van Dyke prosecution also shows that existing fraud and theft-of-government-information statutes could target PM-based schemes even without a dedicated PM insider trading law.³⁰

Note for the news reporters: The definitional dispute is not an abstraction. There is the question of news media responsibility in providing legitimacy to a new sector of businesses whose legal validity and regulatory guardrails are themselves fraught and developing.

A news organization that accepts a data partnership with Kalshi or Polymarket may be accepting revenue from an entity that regulators in other jurisdictions are classifying as a gambling operator. A local U.S. newsroom in Nevada or Arizona that embeds Kalshi odds in its election coverage may be platforming a service that its state courts have found operates illegally.

There are both risks and information here that is not being clearly disclosed to audiences that may become bettors as part of exposure to PM brands on news tickers, or betting odds embedded in reported stories.

Can PMs pressure reporters and affect editorial independence?

Polling methodologies may be contested, and horse-race poll reporting has attracted sustained criticism from media ethicists. But polling does not create direct financial incentives for third parties to pressure individual journalists to alter specific stories. No one loses money because a poll story is worded one way rather than another.

Prediction markets introduce an entirely new dynamic. Because reporting on an event can determine how a PM contract resolves, bettors have a direct financial stake in how journalists describe what happened. The Fabian case (Times of Israel, March 2026) is the clearest documented instance. Emanuel Fabian, the newspaper’s military correspondent, reported that an Iranian ballistic missile had struck an open area near Beit Shemesh on March 10. Over $14 million had been wagered on a Polymarket contract asking whether Iran would strike Israeli soil that day. The contract’s resolution clause excluded intercepted projectiles. Bettors who had taken the “No” position—arguing the missile had been intercepted—stood to lose unless Fabian’s report was changed.²⁴

Fabian received death threats. Fabricated screenshots of forged emails, purportedly from Fabian, were circulated on social media claiming he had agreed to correct the story. A journalist at another outlet was approached by an acquaintance and offered a share of the betting winnings if the colleague could persuade Fabian to change his report. Fabian documented all of this in a first-person account published while the harassment was still ongoing. Polymarket condemned the threats and banned the accounts it could identify, but declined to answer specific accountability questions the Times of Israel put to the company, including whether it had contacted Israeli police and whether it had tools to prevent similar incidents.²⁴

This kind of entanglement of journalists flows from the design of PMs. As Bragues and Aitken have both noted, PM contracts require precise, verifiable resolution criteria. Responsible journalism because it has a verification function, then becomes a resolution mechanism in PMs for public events.²³ Every journalist covering that event could involuntarily be placed into a position of being a financial arbiter for bettors worldwide. The Fabian case also surfaces a second concern: Fabian himself noted that journalists “could easily exploit their knowledge for insider trading on the platform”—the same information asymmetry that makes PM harassment of reporters possible also creates temptation in the other direction.²⁴

ProPublica became the first major U.S. newsroom to respond with a formal policy update, prohibiting its journalists from wager on “the outcome of news events on the prediction markets—regardless of whether or not they are involved in coverage of said event.” This puts a stake into the ethics of journalists-as-bettors. But there are deeper concerns how PMs are going to interfere with legitimate journalism. (See sections below for more.) For instance, when journalism is the resolution mechanism on a PM bet, there is the risk of harm to reporters. There are other gaps, all of which need addressing with newsroom policy.²⁹

Challenges for News Media Professionals and Decision-Makers on Prediction Markets

As the previous section of the primer makes it clear, prediction markets pose serious challenges for news reporters. It ranges from the conflation in public understanding by using betting odds as data sources, to experiencing threats from bettors, to being tempted to wager bets themselves and more. All of these need review for use in a norms-and-standards discussion. The table, below go over two broad dimensions where individual decision-making challenges lie.

Section A of the table is Industry Partnerships, which implicates executives and newsroom leaders. Section B is News Coverage which implicates a much broader group of media professionals including editors, broadcast anchors, producers, reporters, newsletter writers, and more. Each row of the table goes into a type of decision, the target group that is likely to face it, what is at stake, and the ethical issues involved. The ethical issues are surfaced using the Markkula Framework for Ethical Decision Making’s moral lenses.

Decision	Who faces it	What’s at stake	Ethical issues
A—Industry Partnerships Between PM Players and News Companies
Accepting data-ticker partnerships e.g. Kalshi with CNN, CNBC, AP; Polymarket with Dow Jones, Substack	Newsroom executives, business development, editorial leadership	Media brands risk normalizing an industry in the middle of definitional and regulatory dispute with U.S. authorities.36 Polymarket claims “journalism is better when backed by live markets,” but the strengths, weaknesses, and regulatory guardrails for PMs are not yet clear.35 In news partnerships, financial terms are opaque—it is unclear who pays whom and what editorial obligations accompany data access.36	News organizations have a duty of transparency to their audience. (Rights) News organizations dependent on PM revenue risk losing—or being seen to lose—independence to critically cover PM platforms. (Virtue)
Distribution: Legitimizing PMs as news publishers e.g. Google News surfacing Polymarket pages as journalism (Apr. 2026)	Search and social platform decision-makers, SEO teams, digital editors deciding what to link or cite	Polymarket uses “BREAKING” and “JUST IN”-type language on social media.35 Google News briefly classified Polymarket betting pages as newsworthy content before removing them.37 Distribution platforms—search, social, AI assistants—risk conflating PM bets with journalistic stories as sources of information.35	Calling a betting page journalism misrepresents the product to the public and harms epistemic autonomy. Conflation impedes public understanding. (Rights)
B—News Coverage
Embedding prediction market odds in editorial content; using PM data alongside political polling data e.g. digital political desks, broadcast segments, Substack writers	Editors, political reporters, election desks, broadcast producers, newsletter writers	PMs and polls measure categorically different things (see primer section).8,1 Media coverage of opinion polls is already criticized for “horse-race” framing.35 Betting odds carry no proportional representation safeguards—they aggregate financially-weighted belief-probabilities, not plain opinion,7 and their epistemic worth across different event types and bettor markets is unconfirmed. The Eric Swalwell case illustrates the risk: PMs had him at 64% (Polymarket) and leading (Kalshi) while conventional polls showed Republicans ahead—PM prices simply reflected bettor sentiment, which was supplanted when the SF Chronicle investigation broke.38,39	Audiences have a right to know whether the numbers they see reflect representative opinion sampling or capital-weighted financial bets. (Rights) Using PM odds as a polling substitute benefits news brands commercially while producing misleading signals for audiences—a net harm and burden. (Utilitarian, Justice)
Covering PM platforms as a news beat e.g. Kalshi, Polymarket	Beat reporters (Wired, NBC, Nieman Lab), trade press, media critics	Reporters must be careful how they frame betting odds. PM odds are neither facts about an event nor authoritative claims—they are wealth-weighted beliefs of bettors and should be explained to the public as such.35,36,40	Responsibility to accuracy for news audiences. (Rights) Obligation to maintain critical independence from the industries they cover, even when those industries are revenue sources for their employers. (Virtue)
Reporting on events with active open PM bets (contract granularity risk)	General reporting profession, especially political, conflict, and foreign correspondents	PM contract resolution criteria may depend on how news reports describe an event.23 The Fabian case (Times of Israel, March 2026): 90%+ of a $14.1M contract’s volume came after the event as traders bet on the wording of a single journalist’s report.24,42 The market was not predicting the future—bettors were wagering on how a reporter would describe the past. Fabricated emails and death threats followed; someone offered Fabian a cash share of winnings to change the story.24 This flows from the design of PM contracts, which require precise, verifiable resolution criteria.1,23	Reporters’ right to autonomy—language choices must serve accuracy, not financial settlement criteria. (Rights) The character of the journalist and institution is defined by whether it resists this pressure. Fabian’s refusal is the virtue case study. (Virtue) Journalists in conflict zones already face physical threats. Adding financial threat vectors from bettors worldwide compounds harm. (Care Ethics)
Covering local/regional events with thin PM bets	Local news editors, community journalists, hyperlocal digital outlets	Kalshi now lists local-level event contracts—municipal races, state ballot measures, local policy outcomes. Trading volumes are very low, often a few hundred dollars.11 Low-liquidity prices carry no reliable signal: the last recorded trade may be stale, and a single large trader can move the price without any new information entering the market.1 Local newsrooms risk treating these small-volume odds as newsworthy community signals when they reflect only one or two bettors’ positions.30	Reporting a thinly-traded local PM bet as community signal misleads the audience. (Rights) Local news organizations, already financially fragile, may be especially susceptible to PM partnership offers that monetize thin local markets, compromising their independence to decline poor-quality data. (Virtue)
Journalists trading on events they cover	Individual journalists—especially political, financial, and foreign correspondents	Journalists routinely receive information before the public—under embargo, off-the-record, or through source conversations. They are also underpaid, creating financial temptation.35 In PMs (unlike SEC-regulated stock markets), trading on non-public information is not necessarily illegal under CFTC rules30, but may invite action under other federal statutes as the Van Dyke case shows.   Fabian noted that journalists “could easily exploit their knowledge for insider trading on the platform.”24 ProPublica has responded with a ban: “no employee should wager on the outcome of news events on the prediction markets—regardless of whether or not they are involved in coverage of said event.”29	A journalist’s credibility depends on audiences trusting they have no financial stake in the story’s framing or timing. Trading on private information violates that trust categorically. (Virtue) Audiences and sources have a right to expect reporting is not shaped by financial self-interest. (Rights)
Sources placing bets before approaching journalists	More exposed beats: politics, financial, national security	A source holding material non-public information may place PM bets before or during a conversation with a journalist, knowing the story will shift odds. The journalist’s reporting then becomes the financial instrument of the source’s gain.30 There is no clear prohibition on this under CFTC rules.30 In Israel, a military reservist and civilian were indicted for using classified information to bet on the Israel–Iran war—showing state actors have already exploited this gap.24 Journalists cannot reasonably interrogate every source about PM positions, but should research open bets before source conversations and be alert to pressure for language aligned with resolution criteria.27	A source exploiting the journalist-source relationship for private financial gain violates the care relationship it depends on. (Care Ethics) The journalist’s work product—the act of reporting—is used without consent as a financial instrument. (Rights) Alertness to this pattern and preparation before source conversations is part of professional responsibility. (Virtue)

Ethical lenses drawn from: Markkula Center for Applied Ethics, “A Framework for Ethical Decision Making,” scu.edu/ethics.

Acknowledgments

Claude.ai was used to a) search the references and cluster relevant textual sections for manual review b) draw up rough tables using the author’s category vocabulary. This was then recast and written out by author into the sections. It was also used to insert the footnotes, converting from the author’s notings, and structure the references in APA style.

References

1. Bragues, G. (2009). Prediction markets: The practical and normative possibilities for the social production of knowledge. Episteme, 6(1), 91–106. https://doi.org/10.3366/E1742360008000567
2. Toff, B. (2019). Horse-race and game-framed journalism’s effects on turnout, vote choice, and attitudes toward politics. In E. Suhay, B. Grofman, & A. H. Trechsel (Eds.), The Oxford handbook of electoral persuasion. Oxford University Press. https://doi.org/10.1093/oxfordhb/9780190860806.013.24
3. Erikson, R. S., & Wlezien, C. (2012). The timeline of presidential elections: How campaigns do (and do not) matter. University of Chicago Press.
4. Jennings, W., & Wlezien, C. (2018). Election polling errors across time and space. Nature Human Behaviour, 2(4), 276–283. https://doi.org/10.1038/s41562-018-0315-6
5. Journalists’ Resource. (2024, July 10). The consequences of horse race reporting: What the research says. Harvard Kennedy School Shorenstein Center. https://journalistsresource.org/politics-and-government/horse-race-reporting-election/
6. Westwood, S. J., Messing, S., & Lelkes, Y. (2020). Projecting confidence: How the probabilistic horse race confuses and demobilizes the public. Journal of Politics, 82(4), 1530–1544. https://doi.org/10.1086/708682
7. Wolfers, J., & Zitzewitz, E. (2006). Prediction markets in theory and practice. Stanford GSB Research Paper No. 1955. Stanford Graduate School of Business. [See also: Wolfers, J., & Zitzewitz, E. (2004). Prediction markets. Journal of Economic Perspectives, 18(2), 107–126. https://doi.org/10.1257/0895330041371321]
8. Manski, C. F. (2004). Interpreting the predictions of prediction markets. NBER Working Paper No. 10359. National Bureau of Economic Research. https://doi.org/10.3386/w10359
9. American Association for Public Opinion Research (AAPOR). (2024). Standard definitions: Final dispositions of case codes and outcome rates for surveys (10th ed.). AAPOR. https://aapor.org/standards-and-ethics/ [See also: Groves, R. M., Fowler, F. J., Couper, M. P., Lepkowski, J. M., Singer, E., & Tourangeau, R. (2009). Survey methodology (2nd ed.). Wiley.]
10. Pew Research Center. (n.d.). Methods 101: What is a margin of error and how do I use it? Pew Research Center. https://www.pewresearch.org/methods/u-s-survey-research/sampling/
11. Brorsen, B. W., & Fofana, N. F. (2001). Success and failure of agricultural futures contracts. Journal of Agribusiness, 19(1), 129–145.
12. American Association for Public Opinion Research (AAPOR). (2021). An evaluation of 2020 election polls in the United States. AAPOR. https://aapor.org/evaluating-2020-us-election-polls/ [See also: Kennedy, C., & Hartig, H. (2021). Response rates in telephone surveys have resumed their decline. Pew Research Center. https://www.pewresearch.org/methods/2021/01/26/response-rates-in-telephone-surveys-have-resumed-their-decline/]
13. American Association for Public Opinion Research (AAPOR). (2017). An evaluation of 2016 election polls in the United States. AAPOR. https://www.aapor.org/Education-Resources/Reports/An-Evaluation-of-2016-Election-Polls-in-the-U-S.aspx
14. Berg, J., Nelson, F., & Rietz, T. (2003). Accuracy and forecast standard error of prediction markets. University of Iowa working paper. [See also: Berg, J., Gruca, T., & Rietz, T. (2008). Prediction market accuracy in the long run. Mathematical Social Sciences, 56(2), 204–220. https://doi.org/10.1016/j.mathsocsci.2008.04.003]
15. Sunstein, C. R. (2006). Deliberating groups versus prediction markets (or Hayek’s challenge to Habermas). Episteme, 3(3), 192–213. https://doi.org/10.3366/epi.2006.3.3.192
16. Rhode, P. W., & Strumpf, K. S. (2004). Historical presidential betting markets. Journal of Economic Perspectives, 18(2), 127–141. https://doi.org/10.1257/0895330041371277
17. Gruca, T. S., & Rietz, T. A. (2024). Iowa Electronic Markets: Forecasting the 2024 US presidential election. PS: Political Science & Politics. https://doi.org/10.1017/S1049096524000921
18. American Association for Public Opinion Research (AAPOR). (2025 51³Ô¹Ï). AAPOR task force on 2024 pre-election polling: Report. AAPOR. https://aapor.org/wp-content/uploads/2025 51³Ô¹Ï/10/AAPOR-Task-Force-on-2024-Pre-Election-Polling_Report.pdf
19. Krosnick, J. A. (1991). Response strategies for coping with the cognitive demands of attitude measures in surveys. Applied Cognitive Psychology, 5(3), 213–236. https://doi.org/10.1002/acp.2350050305 [The term “satisficing” originates in: Simon, H. A. (1956). Rational choice and the structure of the environment. Psychological Review, 63(2), 129–138. https://doi.org/10.1037/h0042769]
20. Hanson, R. (1999). Decision markets. IEEE Intelligent Systems, 14(3), 16–19. https://doi.org/10.1109/5254.769886
21. Grossman, S. J., & Stiglitz, J. E. (1976). Information and competitive price systems. American Economic Review, 66(2), 246–253.
22. Whitaker, N. (2021). Can you rationally disagree with a prediction market? The Brown University Journal of Philosophy, Politics & Economics, 3(1), 3–16.
23. Aitken, R. (2011). Financializing security: Political prediction markets and the commodification of uncertainty. Security Dialogue, 42(2), 123–141. https://doi.org/10.1177/0967010611399617
24. Fabian, E. (2026, March 16). Gamblers trying to win a bet on Polymarket are vowing to kill me if I don’t rewrite an Iran missile story. Times of Israel. https://www.timesofisrael.com/gamblers-trying-to-win-a-bet-on-polymarket-are-vowing-to-kill-me-if-i-dont-rewrite-an-iran-missile-story/
25. Krosnick, J. A. (1999). Survey research. Annual Review of Psychology, 50, 537–567. https://doi.org/10.1146/annurev.psych.50.1.537
26. U.S. Department of Justice. (2026, April 23). U.S. soldier charged with using classified information to profit from prediction market bets [Press release]. https://www.justice.gov/opa/pr/us-soldier-charged-using-classified-information-profit-prediction-market-bets
27. CNBC. (2026, April 15). Kalshi and Polymarket lobby as insider trading eyed by Congress. CNBC. [See also: Norton Rose Fulbright. (2026, April 20). Prediction markets at a crossroads. https://www.nortonrosefulbright.com/en/knowledge/publications/2026]
28. AP News. (2026, March 23). Kalshi and Polymarket rush to ban insider trading. Associated Press. [See also: Fortune. (2026, March 24). Prediction markets announce guardrails amid congressional pressure.]
29. ProPublica. (2026, April). ProPublica standards and ethics: Prediction markets policy update. ProPublica. https://www.propublica.org/policies [Reported in: Dhanesha, N. (2026, April 15). Prediction markets are breaking the news and becoming their own beat. Nieman Lab. https://www.niemanlab.org/2026/04/prediction-markets-are-breaking-the-news-and-becoming-their-own-beat/]
30. Whitaker, N. (2021).—see ref. 22. [Whitaker’s discussion of insider information in prediction markets and the distinction from SEC-regulated securities markets, pp. 10–12.]
31. Autorité Nationale des Jeux (ANJ). (2024). Polymarket classification as unauthorised gambling [Regulatory statement]. ANJ, France. [Reported in: iGaming Business. (2026, March 19). Can prediction markets crack Europe’s regulatory block? https://igamingbusiness.com/legal-compliance/can-prediction-markets-crack-europes-regulatory-block/; Decrypt. (2026, January 20). Polymarket banned in Portugal and Hungary.]
32. iGaming Business. (2026, March 19). Can prediction markets crack Europe’s regulatory block? iGaming Business. https://igamingbusiness.com/legal-compliance/can-prediction-markets-crack-europes-regulatory-block/
33. The Center Square. (2026, April). State prediction markets regulatory scorecard: 14 of 16 rulings against platforms. The Center Square. [Citing attorney Daniel Wallach; see also: Nevada Independent. (2026, March–April). Nevada Gaming Control Board on Kalshi; Nevada Current. (2026, March). Arizona AG Kris Mayes files criminal charges against Kalshi.]
34. NPR. (2026, April 2). Trump administration sues three states over prediction markets. NPR. https://www.npr.org/2026/04/02/nx-s1-5348912/trump-cftc-sues-states-prediction-markets-kalshi
35. Dhanesha, N. (2026, April 15). Prediction markets are breaking the news and becoming their own beat. Nieman Lab, Harvard University. https://www.niemanlab.org/2026/04/prediction-markets-are-breaking-the-news-and-becoming-their-own-beat/
36. Khatchadourian, D. (2026, April). The problem with binding news and prediction markets. Columbia Journalism Review. https://www.cjr.org/tow_center/the-problem-with-binding-news-and-prediction-markets-polymarket-kalshi-regulation-cftc-insider-trading.php
37. Futurism. (2026, April 11). Google News is now surfacing Polymarket bets. Futurism. https://futurism.com/future-society/google-news-polymarket [See also: Vincent, B. (2026, April 12). Google News briefly showed Polymarket prediction market bets as if they were news. The Verge. https://www.theverge.com/tech/910691/google-news-polymarket-bets-error]
38. Capitol Weekly. (2026, April). Polls vs. prediction markets: Contrasting perspectives on the governor’s race. Capitol Weekly. https://capitolweekly.net/polls-vs-prediction-markets-contrasting-perspectives-on-the-governors-race/
39. DeFi Rate. (2026, April). Swalwell odds plummet: Sex scandal reprices CA governor race. DeFi Rate. https://defirate.com/news/swalwell-odds-plummet-sex-scandal-reprices-ca-governor-race/
40. Dispatch Staff. (2026, March). Prediction markets, current events, and media ethics. The Dispatch. https://thedispatch.com/article/prediction-markets-current-events-media-ethics/
41. Knibbs, K. (2026). Kalshi has been temporarily banned in Nevada. Wired. https://www.wired.com/story/kalshi-temporarily-banned-nevada/
42. Vardi, N. (2026, March 19). How an irrelevant prediction market detail led to death threats. Bloomberg. https://www.bloomberg.com/news/articles/2026-03-19/how-irrelevant-prediction-market-detail-led-to-death-threats