DeFi Rug Pulls in 2026: Inside Crypto’s Exit Scam Machine

In the time it takes to read this article, dozens of new tokens will be minted across Solana, Base, and BNB Chain. A handful will be sincere. Most will not. And a predictable share exist for one reason only: to gather other people’s money into a liquidity pool and then take it. That maneuver, pulling the value out from under buyers who thought they were early, is the rug pull, and in 2026 it has stopped being a cottage crime and become an assembly line.

The numbers that get quoted are slippery, and we will unpack why, but the direction is not in dispute. Blockchain analytics firm Chainalysis estimates that crypto scams and fraud drained as much as $17 billion in 2025, with the average scam payment jumping from $782 to $2,764 in a single year. Rug pulls are a large slice of that total, sitting alongside pig butchering, phishing, and impersonation. What makes them distinct, and worth their own examination, is that the perpetrator is usually the person who built the thing, and the theft is frequently written into the code before the first buyer ever arrives.

This piece walks through what a rug pull actually is, how the mechanics work at the level of individual transactions, the landmark cases that defined the last eighteen months (including a token endorsed by a sitting head of state), the tools that promise to catch rugs before they happen, and why US regulators are still arguing about whether most of this is even illegal.

What a rug pull actually is

The phrase comes from the older idiom of pulling the rug out from under someone. In crypto it has a precise meaning: the people who create or control a token remove the value that backs it, leaving holders with an asset they cannot sell or one that is suddenly worth nothing. The defining trait is not the size of the loss or the speed of the crash. It is agency. A rug pull is an inside job.

That separates it from two things it is often confused with. A hack is an external attacker exploiting a flaw the developers did not intend, the category that dominates the incident reports from firms like CertiK and covers everything from bridge exploits to compromised hardware wallets. A pump-and-dump is coordinated hype followed by insiders selling into it, which can happen without the developers ever touching a smart contract. A rug pull frequently borrows the mechanics of both, but its signature is that the team itself is the exit.

Practitioners split rugs into two families. A hard rug is a single, decisive act: liquidity is yanked, an unlimited supply is minted and dumped, or a hidden function is triggered, and the money is gone in one block. A soft rug is slower and quieter: the team stops developing, sells its allocation over weeks, and lets the token bleed toward zero while insisting everything is fine. Both end in the same place. Only the tempo differs.

The scale of the problem in 2026

Ask how much rug pulls cost investors and you will get a different answer from every source, which is itself instructive. Chainalysis, whose 2026 report put total scam losses near $17 billion, deliberately does not publish a clean rug-pull line item, because its methodology counts the funds operators actually take, not the paper wealth that evaporates when a token collapses. Those are very different figures. When a memecoin with a $4 billion notional market cap goes to zero, almost none of that $4 billion ever existed as withdrawable liquidity.

That gap is why you should be skeptical of the round numbers that circulate. A widely repeated claim that rugs cost $2.8 billion in 2025 does not appear in any primary Chainalysis document; the closest verifiable figure, from security firm Solidus Labs, is a median rug value of just $2,800 per rug, three zeros smaller. Most rugs are tiny. The damage is a function of volume and of the rare mega-rug, not of the typical scam being large.

On the exploit side, CertiK’s Hack3D report tallied over $1.3 billion lost across 344 incidents in the first half of 2026, with wallet compromises alone accounting for $445 million. Those figures capture hacks rather than rugs, but they set the backdrop: the volume of value flowing through permissionless systems is enormous, and only a fraction of the people moving it can read a smart contract. The real driver of rug-pull losses is the sheer number of tokens. Thousands launch every day, and the base rate of fraud among them is extraordinarily high.

Anatomy of a hard rug, transaction by transaction

To see how a hard rug works, follow the liquidity. Most tokens trade on automated market makers, the decentralized exchanges that hold a pool of the new token paired with something valuable like SOL, ETH, or a stablecoin. Buyers deposit the valuable asset and receive the new token, which pushes the price up along a curve. If you want the mechanics of how these pools price assets, our explainer on automated market maker design covers the math. The rug lives in who controls that pool.

When a developer seeds a pool, they receive liquidity provider tokens representing their claim on it. If those tokens are not locked or burned, the developer can redeem them at any moment, withdrawing the paired SOL or ETH and leaving buyers holding a token with nothing on the other side of the trade. That is the classic liquidity pull. The stages usually look like this:

The entire cycle can run in days or in minutes. Because the deciding action is a legitimate contract call (redeeming your own LP tokens is not, by itself, a hack), nothing technically breaks. The system does exactly what it was built to do. That is the uncomfortable core of the rug pull: it is usually not an exploit of the code, it is the intended use of the code by someone who lied about their intentions.

Honeypots, hidden mints, and the soft rug

The liquidity pull is the crude version. The more sophisticated rugs never touch the pool at all, because they make sure buyers can never leave. A honeypot contract lets anyone buy but contains hidden logic, a blacklist, a transfer restriction, or a sell tax set near 100%, that blocks everyone except the creator from selling. The chart looks healthy, the price climbs, and every buyer is trapped. Only the deployer can cash out.

Other rugs are baked into token supply. A hidden or unrenounced mint function lets the developer create new tokens at will, so no matter how much genuine demand arrives, the creator can print an unlimited amount and dump it into the pool. Upgradeable proxy contracts are subtler still: the code you audited on day one can be swapped for malicious code on day thirty, because the owner kept the right to upgrade it. Each of these is a switch the team can flip later, and each is invisible to a buyer who only glances at the price.

Soft rugs use none of these tricks. They rely on patience. Solidus Labs found that 93% of the liquidity pools it studied on Raydium showed soft-rug characteristics, meaning value was drained gradually rather than in one dramatic move. There is no single block to point to, no obvious crime scene, which is exactly what makes soft rugs so hard to prosecute and so easy to dismiss as a project that simply failed.

The memecoin factory: Pump.fun and the 98.6% number

What turned rug pulls from a craft into an industry was the launchpad. Platforms like Pump.fun let anyone create a token in seconds, with no coding, using a bonding curve that seeds liquidity automatically. The friction that once limited how many scams could exist at once simply disappeared. The results are stark. A Solidus Labs study reported by CoinDesk in 2025 found that 98.6% of tokens launched on Pump.fun ended as rug pulls or pump-and-dump schemes. Of more than seven million tokens minted since the platform’s January 2024 launch, only about 97,000 ever held even $1,000 in liquidity.

Pump.fun rejects the framing. A company spokesperson, Troy Gravitt, told CoinDesk that Solidus Labs lacked “a basic understanding of memecoins,” arguing that “98% of memecoins, just like NFTs, tweets, IG posts, trading cards, and most art, are worth little in the long run. That’s precisely the point.” In this view the platform is a neutral marketplace, and worthlessness is the expected outcome of a cultural bet, not evidence of fraud.

The argument has a real edge to it, but it also elides the difference between a token that fails and a token engineered to fail. The next wave complicates the picture further: AI agents that mint, promote, and trade their own tokens with no human in the loop, a frontier we examined in our look at on-chain AI agents and the 2026 reckoning over autonomy and trust. When the launcher is a bot, the question of intent, the very thing that defines a rug pull, gets harder to answer.

The presidential rug: how LIBRA burned a fortune in seven hours

No case did more to drag rug pulls into mainstream politics than LIBRA. On 14 February 2025, Argentine President Javier Milei posted, then deleted, a message promoting the token to his millions of followers. Within roughly forty minutes it had surged more than 2,000%. CoinDesk reported that LIBRA reached a market cap around $4.4 billion before collapsing, and that insiders pocketed roughly $87 million as the price cratered.

Later on-chain analysis showed just how lopsided the outcome was: a small cluster of wallets exited near the top while the overwhelming majority of buyers were left deep underwater, well over a hundred thousand of them. Investigators pointed to executives connected to Kelsier Ventures, notably Hayden Davis, as having arranged the liquidity and the coordinated trades that let insiders sell first. Billions in notional value evaporated in about seven hours, and the shock rippled out into a broad sell-off across the memecoin market.

The fallout crossed borders and is still unfolding in 2026. US investigators became involved because dollars and infrastructure touched American soil; in April 2026, CoinDesk reported that phone records tied Milei more closely to the people behind the token, keeping a criminal probe alive across Argentina, the United States, and Spain. LIBRA matters not because it was typical, it was extraordinary, but because it showed that the rug-pull template scales all the way up to a national government, and that even at that altitude the on-chain evidence is permanent and legible.

Rug or collapse? The Mantra question

Not every catastrophic crash is a rug pull, and the difference is not always knowable from the outside. On 14 April 2025, the OM token from real-world-asset project Mantra fell more than 90% in about an hour, wiping out something on the order of $6 billion in market value as the price dropped from around $6.30 to under $0.50. On its face it looked like a textbook rug: a thin float, insiders, a vertical collapse.

But the story resisted the label. Mantra’s team attributed the crash to reckless forced liquidations by centralized exchanges during thin Sunday-evening liquidity, not to an insider drain. On-chain sleuths noted that wallets linked to the team had moved large amounts of OM to exchanges shortly before the fall, which fueled suspicion, yet subsequent scrutiny produced no proof that the team had orchestrated an exit. The project stayed operational and announced a large token burn to rebuild confidence.

Mantra is worth dwelling on precisely because it is ambiguous. A rug pull requires intent and control, and both can be genuinely hard to establish when a token has a concentrated supply, heavy leverage on centralized venues, and a team that also happens to hold much of the float. The same on-chain footprint can be read as a smoking gun or as ordinary treasury management. That ambiguity is not a footnote; it is the exact seam that both scammers and honest teams operate inside, and it is where regulators and courts do their hardest work.

The slow rug: when the exit is fully disclosed

The most legally durable version of a rug pull is the one that breaks no rules at all. Many tokens allocate large shares of supply to founders, early teams, and venture investors, released on a vesting schedule over months or years. When those cliffs arrive, insiders can sell into whatever demand retail has built, capturing value they were granted at a fraction of the market price. Nothing is hidden. The unlock calendar is often published on day one. And yet the economic effect, insiders exiting onto later buyers, can be indistinguishable from a soft rug.

This is where the vocabulary strains. Calling a disclosed, contractual token unlock a rug pull is unfair to teams that build real products and vest transparently. But retail buyers rarely model the dilution, and promoters rarely emphasize it, so the practical experience of being on the wrong side of a large unlock can feel identical to being rugged. The trust problem is the same one that runs through the rest of decentralized finance, where users increasingly delegate judgment to intermediaries whose incentives they cannot fully see, a dynamic we explored in our analysis of DeFi lending, modularization, and curator risk.

No scanner can flag this pattern, because there is nothing malformed in the contract to detect. The only defense is reading the tokenomics: who owns what, on what schedule, and how much of the float will hit the market while you are still holding. A token can be perfectly audited, fully renounced, and still be a slow-motion transfer of wealth from newcomers to insiders.

Red flags: reading a token before you buy

Most rugs share a family of warning signs, and while none is proof on its own, they compound. The table below groups the signals that experienced traders check before touching a new token.

The single most reliable discipline is boring: assume a brand-new token with an anonymous team and unlocked liquidity is a rug until the evidence says otherwise. That default costs you nothing but a few missed gambles, and it filters out the overwhelming majority of the tokens that will hurt you.

The detection arms race: tools and on-chain forensics

A small industry now exists to automate the checks in that table. Contract scanners parse bytecode for dangerous functions, simulate buys and sells to catch honeypots, and score wallet clusters for insider patterns. None of them is a guarantee, but used together they catch a large share of the crudest rugs before you commit funds.

Every one of these shares the same fundamental limit: it can read code, not intentions. A scanner cannot predict a legally disclosed insider unlock, cannot know that a team plans to abandon a project next month, and cannot judge whether the hype is organic or bought. This is the same lesson that runs through so many security failures, that tooling narrows the attack surface without ever closing it, a theme our Coldcard post-mortem traced through a hardware bug with no attacker at all. The scanner can tell you a token is probably not a honeypot. It cannot tell you the founder is honest.

The SEC, Howey, and the memecoin jurisdiction fight

Here is the part that surprises newcomers: in the United States, it is not obvious that launching and dumping a memecoin is a securities violation at all. In February 2025, the SEC’s Division of Corporation Finance issued a staff statement declaring that most memecoins are not securities, likening them to collectibles and reasoning that they lack the common enterprise and the reasonable expectation of profit from the efforts of others that the Howey test requires.

Commissioner Hester Peirce, long the agency’s most crypto-friendly voice, has framed this as an honest reading of the law rather than a favor to the industry. She has said that “many of the memecoins that are out there probably do not have a home in the SEC under our current set of regulations.” If a token is not a security, the SEC’s disclosure and registration regime simply does not attach, and neither does much of its enforcement power.

Commissioner Caroline Crenshaw dissented in blunt terms. She called the guidance’s value “questionable, except perhaps as a roadmap for crypto enterprises looking to evade oversight by labeling themselves as a meme coin,” arguing that the profit linkage between promoters and buyers can satisfy Howey’s common-enterprise prong. The split matters enormously for rug victims: if the asset was never a security, the most powerful US financial regulator is largely a bystander, and the case falls to fraud statutes, the CFTC, and criminal prosecutors instead. The same classification questions shape which crypto products can reach regulated markets at all, as our guide to how crypto ETFs get approved lays out.

RICO, receivers, and stings: how prosecutors are adapting

If securities law is an uncertain tool, plaintiffs and prosecutors are reaching for others. The marquee civil case is Aguilar v. Baton Corporation, the class action against Pump.fun’s operator in the Southern District of New York. Investors initially sued over losses on volatile memecoins, and the complaint has since been amended. Wolf Popper, one of the firms behind it, describes an expanded racketeering theory naming Pump.fun’s founders and Solana entities, seeking treble damages and even the appointment of a receiver to take control of the operator. Motions to dismiss were pending as of this writing, and the outcome will signal how far RICO law can stretch to cover a permissionless token platform.

On the criminal side, the Department of Justice has shown it can build cases the old-fashioned way, with undercover work. In March 2026, prosecutors in the Northern District of California charged ten people across four market-maker firms in an operation TRM Labs summarized as a sweeping crackdown on crypto market manipulation. The FBI and IRS Criminal Investigation had created their own token to bait the defendants, who allegedly provided wash-trading services; in one firm’s sampled transactions, 99% were self-dealing between linked wallets. It was not a rug pull in the strict sense, but it targeted the same machinery, the fake volume and manufactured liquidity that make a worthless token look investable.

The throughline is improvisation. With no purpose-built statute for on-chain fraud, US authorities are retrofitting racketeering law, wire-fraud charges, commodities authority, and undercover stings onto a problem that moves faster than any of them. Enforcement lands, but it lands late, and usually only on the largest or most brazen operators.

Gaming rugs, governance rugs, and the AI twist

Rug pulls are not confined to standalone memecoins. Blockchain gaming has its own version: a studio launches a token and NFTs tied to a promised game, collects the mint proceeds and early demand, then quietly abandons development, leaving holders with assets for a product that never ships. Because the token often does carry a nominal roadmap and a Discord full of believers, these soft rugs can drift for months before the community accepts what happened. The pattern is common enough that seasoned players treat an unreleased game with a live, tradable token as a warning rather than a feature.

Decentralized governance introduced a stranger variant, the rug by vote. If an attacker can accumulate enough governance tokens, they can pass a proposal that legitimately transfers a protocol’s treasury to themselves, using the system’s own rules against it. It looks less like theft than like a hostile takeover with extra steps, and it exploits the gap between what a vote is technically allowed to do and what its designers assumed anyone would ever want to do.

The newest wrinkle is autonomy. As AI agents begin to issue and manage tokens on their own, the human intent that anchors the very definition of a rug pull becomes diffuse. If an agent dumps a token it created, following an objective a developer set loosely months earlier, who exactly pulled the rug? The law has no answer yet, and the pace of deployment is not waiting for one.

Can rug pulls be engineered out, and what to do if you are hit

The ecosystem has developed conventions meant to signal safety: locking liquidity for a fixed term, burning LP tokens outright, renouncing contract ownership so no one can mint or pause, and holding treasuries in multisig wallets. Each raises the cost of an obvious rug. None is decisive. Locks expire. Renouncements can be theater layered over a proxy that still holds power. And an audit, the gold standard many buyers look for, certifies that code does what it says, not that the people behind it are honest. Plenty of audited protocols have been drained regardless, because a review of the code says nothing about the character of the team that deploys it.

The honest conclusion is that rug pulls cannot be fully patched out of a system whose defining feature is that anyone can issue anything to anyone without permission. They are a cost of the design space, mitigated by tooling, reputation, and slow legal pressure, not eliminated by any of them. That places the burden on individuals to an uncomfortable degree.

If you are rugged, act quickly but hold realistic expectations. Record every transaction hash and wallet address while the trail is fresh. Report the theft to the FBI’s Internet Crime Complaint Center and, if the token was promoted as an investment, to the SEC; both aggregate complaints that can seed larger cases. Blockchain analytics firms and volunteer investigators sometimes trace and freeze funds, especially when stablecoins are involved and an issuer can blacklist an address, but for the typical small rug the money is gone through a mixer within hours and recovery is unlikely. The best protection remains the cheapest: skepticism before you send, not forensics after.

Frequently Asked Questions

What is a rug pull in crypto?

A rug pull is a scam in which the people who create or control a token remove its value, leaving holders unable to sell or with a worthless asset. The defining feature is that the perpetrator is an insider, the developer or a controlling party, rather than an outside hacker. It can happen in one transaction (a hard rug) or gradually over weeks (a soft rug).

How can I tell if a token is a rug pull before buying?

Check whether the liquidity is locked or burned, whether the team has renounced owner privileges like minting and blacklisting, how concentrated the holders are, and whether you can actually sell by running a honeypot simulation. Anonymous teams, unlocked liquidity, and manufactured hype are the strongest warning signs. Tools such as RugCheck, GoPlus, and Token Sniffer automate many of these checks, though none can detect a dishonest founder or a disclosed insider unlock.

Are rug pulls illegal in the United States?

It depends on how the token is classified. The SEC’s staff has said most memecoins are not securities, which limits securities-law enforcement, but outright fraud, wire fraud, and racketeering statutes can still apply, and the CFTC and DOJ have pursued cases. Prosecutors are increasingly using RICO claims and undercover operations because there is no single statute written specifically for on-chain fraud.

Can you get your money back after a rug pull?

Usually not, though it is not impossible. If you act fast, record the transaction hashes, and report to the FBI’s Internet Crime Complaint Center and the SEC, analytics firms occasionally trace and freeze funds, especially when a stablecoin issuer can blacklist an address. For the typical small rug, the funds move through a mixer within hours and recovery is unlikely.

What is the difference between a rug pull and a crypto hack?

A hack is an external attacker exploiting a flaw the developers did not intend. A rug pull is an inside job: the team itself takes the money, often using functions deliberately built into the token, such as an unlocked liquidity pool or a hidden mint. The mechanics can overlap, but intent and control are what set a rug pull apart.

By Nathan Cole, senior editor covering security and exploits at HOGE Wire.