Someone Rated Your Grandmother's Facebook Post False
Somewhere in the last several years, a peculiar new authority entered public life. It did not run for office. Nobody voted for it. It has no constituents, no term limits, and no process by which it can be removed. And yet it now decides, on a rolling and largely invisible basis, which claims are true enough to circulate and which claims get a small red flag attached that quietly throttles their reach, buries them in search, or triggers an account-level strike.
The fact-checking industry positions itself as neutral. Independent. A kind of referee standing above the fray, calling balls and strikes with no stake in who wins. I want to take that claim seriously enough to actually check it, because as far as I can tell, nobody rating claims for a living has ever rated their own institution's neutrality, and the omission makes it suspect.
The earlier pieces in this series (1, 2, 3, 4, 5, 6, 7, 8, 9) covered how platforms suppress reach, shadowban accounts, and manufacture Overton windows through algorithmic sorting and siloing. This one covers the specific mechanism that triggers a lot of that suppression in the first place. It’s a third-party rating system whose funding sources rarely make it into the byline.
What a “Fact-Check” Actually Is
A fact-check, in the modern platform sense, functions as a label. For nearly a decade, Facebook and Instagram contracted with organizations certified under the International Fact-Checking Network to review flagged content and assign a rating, a system Meta ended in the United States in January 2025 in favor of a crowd-sourced Community Notes model, while keeping the third-party system running in roughly 119 other countries.
Other platforms still use versions of the same mechanism. Content rated false gets its distribution reduced, sometimes (by the platform's own account) by as much as eighty percent. The person who posted it is not notified of the specific reasoning in any detail that would allow a meaningful appeal. The rating simply attaches, and the algorithm does the rest.
Read that method again slowly: A private organization, with a funding structure of its own, rates a claim, and that rating becomes the trigger for algorithmic reach suppression identical in effect to the shadowbanning covered earlier in this series. The fact-check functions as an enforcement mechanism, not merely a piece of information sitting quietly on a page.
The IFCN itself is a project of the Poynter Institute, and its signatories include PolitiFact, Snopes, Lead Stories, Check Your Fact, and dozens of others operating in nearly every country with functioning internet infrastructure. Certification requires adherence to a code of principles that includes, on paper, nonpartisanship and open funding disclosure. The disclosure part is where things get interesting, and where we can see that the “nonpartisan” claim requires a certain level of scrutiny.
The Funders
PolitiFact is a project of the Poynter Institute, which has received funding from the Bill and Melinda Gates Foundation, the Charles Koch Institute, Facebook itself, Craig Newmark Philanthropies, and the Omidyar Network, among others. I think it's worth naming these plainly rather than gesturing at “big donors,” because the specific identity of the funder matters for the specific topic being rated.
The Gates Foundation funds global health initiatives, vaccine distribution programs, and agricultural policy on a massive scale. It also funded, both directly and through intermediary grants, the fact-checking apparatus that later rated claims about the foundation's own financial interests.
Columbia Journalism Review documented that PolitiFact and USA Today, both Gates Foundation grant recipients, used their platforms to rate as false the claim that the foundation held financial stakes in COVID vaccine and treatment developers, in a case where CJR found the foundation's own tax filings showed investments in companies including Gilead and CureVac.
Poynter's senior vice president Kelly McBride confirmed the Gates money was passed through to fact-checking partners including Africa Check. She said she was “absolutely confident” no bias emerged from the arrangement and then acknowledged she had not reviewed the work herself.
I found this while chasing a footnote on an unrelated story, and I spent a full day double-checking before I trusted what I was looking at. Read that again if you need to. A foundation funded the health policy. The same foundation, several steps removed, funded the organization rating a claim about the foundation's own financial interest in that policy. Nobody hid this. It just was never going to be the headline.
NewsGuard, which functions less as a claim-by-claim fact-checker and more as a site-wide credibility rating service used by advertisers and platforms to determine ad placement, launched in 2018 with Publicis Groupe as its lead investor among an eighteen-member funding group. Publicis is one of the largest advertising conglomerates on earth, representing pharmaceutical clients, including Pfizer. NewsGuard has separately held a $750,000 contract with the Department of Defense and won a Pentagon-State Department competition worth $25,000 for a pilot program supporting the Cyber National Mission Force, according to NewsGuard's own press materials. A constitutional law expert representing Consortium News in its lawsuit against NewsGuard called the arrangement an “inherent conflict of interest.” A private company, funded partly by an advertising conglomerate with pharmaceutical clients and partly by a defense contract, assigns credibility scores that determine which outlets get monetized and which get starved of ad revenue.
The State Department's Global Engagement Center directly funded NewsGuard, the Global Disinformation Index, and the International Fact-Checking Network's parent umbrella through the National Endowment for Democracy. This is according to a House Foreign Affairs Committee investigation and a lawsuit that produced an inspector general finding that the GEC had “funded and supported” a censorship-technology repository without properly disclosing it.
Here’s where the funding went. Disinfo Cloud, the GEC platform at the center of the lawsuit, held itself out as a repository of more than 365 tools for identifying disinformation. It cataloged what it flagged as disinformation—and then retweeted content from NewsGuard and the Global Disinformation Index.
In April 2021, its own public digest promoted NewsGuard's “Responsible Advertising for News Segment,” a product that was built to help advertisers avoid the sites that NewsGuard had already flagged. Two of the outlets NewsGuard flagged were The Federalist and The Daily Wire—they were each ranked "unreliable" in NewsGuard's own scoring. Those are the same two outlets that later sued the government agency funding the tool doing the ranking.
The conflict of interest was also apparent when following the money. NewsGuard took a $25,000 GEC grant in 2020 to build a database for a joint GEC-Pentagon program designed to “pre-bunk” foreign disinformation before it spread. The Global Disinformation Index took a $100,000 GEC grant in 2021, which was routed through a middleman called Park Advisors instead of being paid directly.
Park Advisors also ran Disinfo Cloud itself on a separate $3 million grant, one that the State Department's own inspector general found that the agency had never disclosed.
Congress declined to renew the GEC's authorization at the end of 2024, and the office closed. A government agency spent eight years funding organizations whose stated purpose was to identify which information is true, and it took a federal lawsuit and an inspector general's report to establish what that funding had actually been paying for.
Meta operated a third-party fact-checking program from 2016 through January 2025, paying the participating organizations directly. Bill Adair, PolitiFact's founder and a co-founder of the IFCN, said Meta had provided “millions of dollars” to fact-checkers worldwide. Mark Zuckerberg ended the U.S. program in January 2025, replacing it with a crowd-sourced Community Notes system, and Meta's own former IFCN director predicted fewer fact-checking organizations would survive the loss of that funding. The arrangement continues today in roughly 119 other countries.
For nine years, the company being fact-checked was also the client paying for the fact-check on its own platform. That is about as clean an illustration of a conflict of interest as the modern information ecosystem has produced, and it operated in full public view because almost nobody looks at the “About” page long enough to notice.
Part 9 of this series named Operation Mockingbird directly: the documented mid-twentieth-century CIA program to place assets inside major American news organizations, confirmed under oath in the Church Committee hearings of 1975. The mechanism Mockingbird demonstrated was simple. A small number of coordinated institutional voices, if positioned correctly, can produce the appearance of a broad and independent consensus.
That mechanism worked in 1975 without any fact-checking apparatus at all. It works considerably more efficiently now that one exists, because a fact-check does something a planted news story could never quite do on its own. It retroactively labels the dissenting position as false, on a platform that then physically restricts how far that position can travel. The coordinated narrative you are competing with has been granted a quasi-official verdict against you, enforced by an algorithm, funded in part by the entities whose claims happen to survive the rating process intact.
The Failure of Imagination: Why People Think This Is Neutral
“Fact-checkers are independent journalists doing their jobs.” Plenty of individual fact-checkers are trying to get things right. That is a separate question from who funds the organization they work for. The funding structure decides which topics get resourced for scrutiny and which get waved through, regardless of what any individual fact-checker intends.
“If the funding were a problem, someone would have exposed it by now.” It has been exposed, repeatedly—in congressional testimony, in the Gates Foundation's own public grant database, in Meta's own vendor payment disclosures. Exposure was never the hard part. A false rating gets pushed out to millions of people by the same algorithm that suppressed the original post. The story about who funded that rating reaches almost nobody, because nothing is built to amplify it.
“Rating something false because it is false isn't bias.” Correct, when the rating tracks the actual state of evidence. But a claim can be technically imprecise, actively contested among credentialed researchers, or simply ahead of where the evidence eventually lands, and it still gets the exact same “false” label as a claim that is flatly fabricated. One rating system, one output, applied to a wide range of very different situations. And the funder's interests line up a little too neatly with which situations get flattened into being “false” and which get treated with more care.
What You Can Do About It
Treat every fact-check as a claim requiring its own verification, not as a terminal ruling. This is the only intellectually honest response to a rating system with an undisclosed financial relationship to at least some of the claims it rates. It sounds exhausting, and it certainly can be. But for the sake of accuracy and honesty, you should do it anyway.
Look up the funding disclosure for the specific fact-checking organization behind a rating before treating that rating as dispositive. Most of the major players publish some version of a funders list, usually buried several clicks deep on an “About” or “Ethics” page. The Poynter Institute, Snopes, and NewsGuard all have public funding pages if you search for them directly rather than through the fact-check article itself, which rarely links to its own institutional disclosures.
Use Media Bias/Fact Check and Ad Fontes Media as a secondary layer. Note that these are themselves organizations with their own methodology and their own blind spots, not neutral oracles either. The goal is triangulation across sources with different institutional incentives, not finding the one source that finally gets to be trusted absolutely.
Read the original claim yourself before reading the fact-check of the claim. This seems obvious, but it is skipped constantly. A fact-check summarizes and reframes the original assertion before rating it, and that summarization step is where a great deal of the actual distortion happens. Track down the primary statement, video, or document, form your own read, and then compare it to how the fact-check characterized it.
Run a basic source trace on any viral claim before accepting either the claim or its debunking. A reverse image search through TinEye or Google Images. A Wayback Machine check on whether an article's language was quietly edited after publication. A search of the original source's other published work to see whether the claim fits a documented pattern or could be seen as an outlier.
Support and read independent journalists and researchers. Look for ones who publish their funding sources voluntarily and operate outside the certified fact-checking apparatus entirely. Their accuracy still requires the same scrutiny as anyone else's. Their positioning, at least, is disclosed up front, which is the baseline the certified apparatus has struggled to meet.
The Referee Has a Team
A referee who is being paid by one of the teams on the field is not a referee anymore. He is a player wearing a different uniform, and the call he makes in the final minute deserves exactly the scrutiny you would give any player with something to gain from the outcome.
The fact-checking apparatus asked for your trust on the premise of neutrality. The funding record does not support the premise. Read the ratings. Read who paid for them. Draw your own conclusions, which is, after all, the entire point of an open information environment that nobody gets to referee for you.
