Skip to main content

The Win That Never Cleared

Manipulation Breakdowns · 9 min read · By D0

Fifty Losers, One Fake Winner

In January 2026, a college student named George Makihara posted a TikTok video showing a $100,000 win on a bet that Donald Trump would say the word “McDonald’s” during a public appearance. The video looked like a screen recording: a trading interface, a green number, the unmistakable satisfaction of a wager paid off.

More than fifty real people placed that exact bet on the real platform. Public blockchain records show every one of them lost.

Makihara’s win was never placed. It couldn’t have been. The bet resolved the other way. The video was filmed on a cloned website built to look like Polymarket, the prediction market platform that paid him to make it.

What the Journal Found

Polymarket ran a marketing campaign, reported by the Wall Street Journal in early July 2026, that turned “watch me win” into an assembly line. The Journal reviewed more than 1,100 videos posted by ten creators between December 2025 and mid-May 2026 — a sample drawn from a wider campaign involving dozens of influencers. Nearly 70 percent of those videos depicted bets totaling roughly $1.9 million. None of those wagers were placed on Polymarket’s actual platform. In another 118 videos, creators celebrated close to $900,000 in supposed winnings. Public trading records show the same positions, had they been real, would have lost more than $166,000.

The gap between the fiction and the ledger isn’t a rounding error. It’s a full inversion: the platform’s marketing arm manufactured a version of reality where its product made people rich, while the verifiable record of the product’s actual behavior shows the opposite.

The videos accumulated more than 140 million views across TikTok, YouTube, and Instagram.

The Clone in the Machine

The mechanics were straightforward once exposed. Creators filmed their “trades” on dummy websites — the Journal identified at least one, registered as poiymarket.com, swapping a lowercase “l” for an “i” — built to mirror Polymarket’s real interface closely enough to survive a fifteen-second TikTok clip. Small tells distinguished the clones from the live site: altered buttons, missing market data, interfaces that wouldn’t withstand close inspection but didn’t need to. Nobody watching a TikTok pauses to check whether the odds ladder on screen matches the live order book.

The clone site did one job: produce footage. The footage did the rest. A screen recording of a winning trade doesn’t announce itself as fabricated. It looks exactly like what it’s imitating, because that’s the entire design brief.

Polymarket hired a marketing firm called Virality to run the distribution side. Virality operated through a network the Journal describes as “clippers” — many of them reportedly teenagers based in Asia, running multiple accounts, paid according to how much of their audience skewed American. Creators who appeared on camera were paid $2,000 to $3,000 a month and told not to disclose the arrangement. Some added “@polymarket partner” to their bios only after the Journal started asking questions.

Why the Ledger Is the Point

Prediction markets sell themselves on a specific promise: everything settles on a public, immutable ledger. No backroom, no house thumb on the scale, no dispute about who won what — the blockchain says so, permanently and checkably. That transparency is Polymarket’s differentiator against a traditional sportsbook, where the house’s numbers are the only numbers.

This is what makes the marketing failure genuinely interesting, not just another paid-influencer scandal. The fabricated wins didn’t try to fake the ledger. They bypassed it entirely, staging the win on a set that never touched the chain. The verifiable record Polymarket depends on for legitimacy was simply routed around, off to the side, where a viewer would never think to look for it because nothing on screen suggested there was anywhere else to look.

And the ledger is exactly what caught it. The Journal didn’t need a whistleblower or a leaked internal memo. It cross-referenced the bets shown in the videos against Polymarket’s own public trading data — the same transparent record the platform advertises as its core feature — and found no matching trades, or matching trades with the opposite outcome. The tool built to make Polymarket trustworthy is the tool that proved a chunk of its marketing wasn’t.

That’s the mechanism worth sitting with: verifiability is not the same as verification. A ledger that anyone can check is not a ledger everyone does check. Polymarket’s transparency was real the entire time the deception was running. It sat there, public, checkable, correct — and did nothing to stop 140 million views of fiction until a reporter decided to look.

The Disclosure That Wasn’t

Undisclosed paid promotion is not a new problem. What’s notable here is the infrastructure built specifically to keep it undisclosed — a marketing firm as an intermediary layer, a clipper network paid by audience geography rather than by content quality, and explicit instructions to creators not to mention the arrangement.

Each layer does a job:

  • The intermediary firm creates distance between the brand and the instruction to deceive. Polymarket didn’t have to tell Makihara to hide the deal; Virality did, or the clipper network did, or nobody had to say it explicitly because the incentive structure made the choice for them.
  • The clipper network provides scale without accountability. A teenager running six accounts isn’t building a personal brand that has to survive scrutiny — they’re a distribution node, paid on throughput.
  • The non-disclosure instruction preserves the illusion that these are organic testimonials rather than paid advertisements, which matters because audiences weight the two completely differently. A friend’s TikTok about a big win reads as evidence. An advertisement for a betting platform reads as an advertisement.

The entire stack exists to make paid speech look like unpaid speech, at a volume no single brand deal could achieve and with a chain of intermediaries long enough that no one person has to own the instruction to lie.

The Psychology Being Targeted

Fake wins work on a specific, well-understood lever: social proof under uncertainty. Prediction markets, like sports betting and crypto trading, ask users to act under genuine uncertainty about outcomes. The instinctive shortcut for reducing that uncertainty is to look at what other people are doing and how it’s working out for them. A platform full of visible winners is a platform that looks solved — other people already figured out how to win here, so the uncertainty that would otherwise stop you from betting real money gets absorbed by their apparent success.

This is the same tactic that powers fake product reviews, inflated follower counts, and manufactured grassroots political support — manufactured consensus, aimed at collapsing an audience’s uncertainty faster than genuine evidence would allow. The only thing that changes across contexts is the currency. In an election, the manufactured consensus says a policy is unpopular. In a betting market, it says the platform pays out. The tactic is identical: fabricate the thing people would otherwise use real evidence to evaluate, and let the fabrication do the evaluating for them.

The Makihara video is a clean illustration because it’s falsifiable in a way political disinformation often isn’t. Nobody can produce a ledger proving what percentage of a population “really” supports a policy. Polymarket’s own blockchain can, and did, prove exactly what happened to the fifty-plus people who made the real version of his fake bet. They lost. He didn’t. The only difference between them is that his loss was never real.

What Regulators Are Doing About It

On June 26, 2026, Senators John Curtis of Utah and Adam Schiff of California sent a bipartisan letter to the Commodity Futures Trading Commission requesting a formal investigation, calling the reported conduct “deeply troubling” and asking for a response by July 10. The CFTC’s interest in Polymarket predates this campaign: the agency banned the platform from operating in the United States in 2022 following a settlement over running an unregistered options exchange, then granted conditional reentry in 2025 through an invite-only, iPhone-exclusive regulated product — a narrow doorway compared to the offshore platform’s global reach and trading volume.

Polymarket’s response, so far, is an internal audit: the company says it is conducting “a comprehensive audit of active promotional content” to verify compliance with its own standards and “applicable regulatory and legal disclosure requirements.” That’s a company checking its own marketing after a newspaper checked it first — the same post-hoc verification pattern that shows up whenever an institution’s compliance process activates only once external reporting has already done the work.

What a Transparent Ledger Doesn’t Guarantee

The uncomfortable finding here isn’t that a betting platform paid influencers — that’s an old story with a new coat of paint. It’s that “verifiable by design” and “verified in practice” are different claims, and audiences, regulators, and apparently the platform itself treated the first as if it satisfied the second.

A public blockchain removes the need to trust a company’s internal numbers. It does not remove the need for someone, at some point, to actually pull the numbers and check them against the claim being made. For five months and 140 million views, nobody did. The mechanism that could have caught the fabrication in real time — cross-referencing on-screen claims against public trade data — is exactly the mechanism a reporter eventually applied after the fact, at which point the views had already accumulated and the audience already formed whatever impression of Polymarket those fake wins were designed to create.

Anthony Clewis found out his name was on a fabricated public comment because a reporter called him. The fifty-plus people who lost real money on the same bet Makihara faked winning will likely never know their loss became someone else’s marketing material. The ledger recorded their loss faithfully, permanently, exactly as designed. It just took a newspaper to read it.

This article is part of Decipon’s Manipulation Breakdowns series, examining specific influence operations through the Influence Tactics Protocol.

Sources: