Rug Pull Statistics 2026: Losses, Detection and Recovery

Crypto scams received at least $14 billion on-chain during 2025. None of that total is broken out as a rug pull, and no agency counts them separately. Rug pull statistics have to be assembled from adjacent categories: the FBI’s Internet Crime Complaint Center reports investment fraud and cryptocurrency as a complaint descriptor, while Chainalysis reports scam receipts. Every published rug pull total is an inference from a wider aggregate, which is why estimates disagree by billions.

Below are the figures those measurement systems do record: reported losses by crime type, complaint volumes, and the on-chain research measuring how fast new tokens collapse. Detection coverage gaps, fund-recovery rates, and enforcement outcomes follow from there.

Key Takeaways

• The FBI logged $8,648,617,756 in reported investment fraud losses in 2025, the single largest complaint-loss category and the bucket most rug pull reports land in.
• Complaints carrying the cryptocurrency descriptor totalled $11,366,669,732, with nearly 49% of all scam-related losses coming from investment fraud alone.
• The average scam payment rose from $782 in 2024 to $2,764 in 2025, a growth of 253%, so fewer victims now account for far more value.
• A 2026 study assembled 6.4 million Solana tokens over 7 months and found that a vast majority exhibit rug pull characteristics within one hour of launch.
• Detection tooling reaches only part of the problem: 14 detection tools identify 25 of the 34 documented root causes, a coverage of 73.5%.
• The IC3 Recovery Asset Team froze $679,013,183 of $1,163,919,846 in attempted theft, a 58% success rate, but that process depends on bank accounts rather than liquidity pools.

Editor’s Choice

• Illicit crypto addresses received at least $154 billion in 2025, a 162% increase year over year.
• Chainalysis projects the 2025 scam figure could exceed $17 billion as more wallet addresses are identified.
• Cryptocurrency investment fraud alone cost Americans $7.2 billion in reported losses.
• The IC3 received 1,008,597 total complaints, up from 859,532 in 2024.
• Investment fraud generated 72,984 complaints in 2025 against 39,570 in 2023.
• Uniswap research on Ethereum labeled 27,588 tokens for the foundational rug pull dataset, using data collected until 3 September 2021.
• The SafeMoon chief executive was sentenced to 100 months in prison for defrauding investors.

Rug Pull Statistics Inside 2025 Crypto Fraud Losses

• Chainalysis put on-chain scam receipts at at least $14 billion for 2025, up from the $9.9 billion first reported for 2024.
• That 2024 figure reached $12 billion at recalculation, showing how much these estimates move after publication.
• The FBI counted approximately 453,000 cyber-enabled fraud complaints with reported losses exceeding $17.7 billion.
• The cryptocurrency descriptor carried $11,366,669,732 in reported losses across the year.
• Complainants over 60 reported approximately $7.7 billion in losses, up 37% from 2024.

Source: Chainalysis 2026 Crypto Crime Report; FBI Internet Crime Complaint Center 2025 Annual Report

Two measurement systems produce two different pictures. Chainalysis traces value arriving at wallet addresses it has tied to scams, which catches money victims never reported.

Victim complaints drive the FBI figure instead, so it captures losses that never touched a public blockchain. Reading either as the rug pull total is the error running through most coverage of scams in cryptocurrency.

Descriptors give a second cut of the same file, and it is the closest federal reporting gets to isolating crypto-denominated harm.

Reported Fraud Losses by Crime Type

• Investment fraud led every category at $8,648,617,756, more than double the next entry.
• Business email compromise followed at $3,046,598,558.
• Tech and customer support fraud reached $2,134,675,818.
• Personal data breach complaints accounted for $1,314,923,988.
• Confidence and romance fraud cost victims $929,287,469.
• Government impersonation reached $797,943,193, a category Chainalysis separately links to sharp on-chain growth.

Investment is where a rug pull victim’s complaint most plausibly lands. A token buyer who watched liquidity vanish describes an investment that failed, not a hacking incident, so the complaint is categorized accordingly. That routing is why the investment total is not a rug pull total.

Don't chase the news. Let us curate it.

You get one weekly briefing with only the stories that matter. If the market is quiet, we skip it.

Email address

Get the Weekly Briefing - it's free!

✅ Join readers from Visa, Mastercard, Vanguard, and the FDIC.

Recent Developments

• January 2026: Chainalysis estimated $17 billion stolen in crypto scams and fraud in 2025 in its 2026 Crypto Crime Report.
• January 2026: The same report recorded impersonation scams growing more than 1400% compared to 2024, with average payment severity increasing by over 600%.
• February 2026: A Brooklyn court ordered him to forfeit approximately $7.5 million at the SafeMoon sentencing.
• April 2026: The FBI reported cyber-enabled crimes defrauded Americans of nearly $21 billion in its 2025 Internet Crime Report.
• April 2026: Operation Level Up had surpassed 8,000 total victims notified and reduced losses by more than $500 million.
• August 2026: A Solana preprint assembled a dataset of 6.4 million tokens over 7 months.

Fraud Complaint Volume by Crime Type

• Phishing and spoofing produced 191,561 complaints, the largest single category by volume.
• Extortion followed with 89,129 complaints.
• Investment fraud generated 72,984 complaints despite carrying by far the largest loss total.
• The cryptocurrency descriptor appeared in 181,565 complaints, spanning many crime types at once.
• Personal data breach complaints reached 67,456, and non-payment or non-delivery reached 56,478.
• Ransomware, by contrast, produced only 3,611 complaints.

Ratios between the two federal tables carry the useful signal. Investment produced 72,984 complaints against 191,561 for phishing, while investment losses reached $8,648,617,756 against $215,843,126 for phishing. Low-frequency, high-severity is the shape a token collapse produces, as our crypto fraud and security data shows.

By the numbers: Per the FBI’s Internet Crime Complaint Center, investment fraud produced $8,648,617,756 in reported losses during 2025 across 72,984 complaints. Phishing generated 191,561 complaints but only $215,843,126 in losses, a per-complaint gap of more than two orders of magnitude.

How Many New Tokens Show Rug Pull Behavior

• The largest study to date assembled 6.4 million Solana tokens over 7 months, drawn from PumpFun and Raydium.
• Its market analysis found a vast majority of these memecoins exhibit rug pull characteristics within one hour of launch.
• Earlier Ethereum work labeled 27,588 tokens using Uniswap data collected until 3 September 2021.
• The rug pull systematization found existing open datasets document 2,448 instances across the published literature.
• The same authors constructed a replacement dataset containing 2,360 instances with broader cause coverage.

Source: arXiv preprints 2608.20271 and 2403.16082; MDPI Mathematics 2022

Scale is the difference between the two chains, and it changes what any percentage in rug pull statistics means. Ethereum research counted tokens in the tens of thousands over roughly two years; Solana research counted millions in seven months. One percentage across both would flatten a launch-economics difference our memecoins coverage has documented since launchpads arrived.

The Rug Pull Detection Window

• Solana memecoin rug pulls are predominantly driven by liquidity manipulation and social dynamics rather than contract backdoors.
• Gradient boosting models detect potential rug pulls using only the first 5 minutes of trading data.
• The 2022 Ethereum model reached an accuracy of 0.9936, recall of 0.9540, and precision of 0.9838 when a token could be evaluated at any block.
• A fixed evaluation time dropped that to an accuracy of 0.992, recall of 0.784, and precision of 0.869.
• Cross-platform testing between PumpFun and Raydium showed that multi-source data fusion significantly mitigates domain shift.

Recall is where the gap opens. Accuracy barely moves between the two scenarios. Recall falls by roughly a sixth once the model has to decide at a fixed moment instead of picking its vantage point. Detection is easier in hindsight than in the window where it helps, and four years compressed that window to minutes.

How is a rug pull executed on-chain?

A simple rug pull runs in three steps, per the Uniswap research. The developer creates an ERC-20 token and interacts with the Uniswap V2 Factory to create a new trading pair, then investors execute swap transactions on that pair, and the developer then activates the function removeLiquidity. Holders are left without an exit market, as how a crypto rug pull works, as set out in full.

How do you spot a rug pull in crypto?

Research points to timing and liquidity rather than any single visible feature. The Solana work shows that a vast majority of studied tokens show rug pull characteristics within one hour of launch and that trading-data models work despite the absence of code-level features. The observable signals are behavioral, and the reading window is short. CoinLaw’s guidance on spotting a crypto scam covers the wider pattern set.

Where Rug Pull Detection Tools Fall Short

• The systematization documents a taxonomy inclusive of 34 root causes drawn from academic and industry sources.
• Existing open datasets address only 7 of the 34 root causes, amounting to a mere 20% coverage.
• The replacement dataset expands that coverage to 54%.
• Examination of 14 detection tools showed they can identify 25 of the 34 root causes, achieving a coverage of 73.5%.
• Six categories were added from industry sources alone: burn, hidden owner, ownership transfer, unverified contract, external call, and fake LP lock.

Worth noting: Detection coverage is uneven by design. Per the SoK rug pull analysis, 14 detection tools identify 25 of the 34 root causes, a coverage of 73.5%, while existing datasets address only 7 of the 34 root causes, a mere 20% coverage. The evidence base used to train detectors is narrower than the detectors themselves.

Tooling built to read Ethereum bytecode does not transfer cleanly elsewhere, and our Ethereum and Polygon coverage tracks how differently each ecosystem’s standards evolved.

What are the red flags of a rug pull?

Research frames this as a coverage question, not a checklist. Of the 34 documented root causes, the examined tools reach 25, so nine sit outside every tool. Those nine include fake LP lock, hidden fee, destroy token, and fake money transfer. A token passing an automated scan has cleared what tooling models, not what researchers catalogued.

Scam Receipts and Average Payment Size

• Chainalysis first reported $9.9 billion for 2024 before revising the figure upward.
• That estimate reached $12 billion at recalculation as more addresses were attributed.
• The 2025 first estimate came in at at least $14 billion on-chain.
• Applying the same revision pattern, the firm projects the 2025 figure could exceed $17 billion, since annual estimates grow by an average of 24% between reporting periods.
• Average payment size moved from $782 in 2024 to $2,764 in 2025.

Revision patterns are why no single year of rug pull statistics should be treated as settled. Each year’s number climbs after publication as attribution work continues, so the freshest estimate is always the most understated one.

Three-Year Change in Fraud Complaint Volume

• Investment complaints climbed from 39,570 in 2023 to 47,919 in 2024 and 72,984 in 2025.
• Government impersonation more than doubled, from 14,190 to 32,424 over the same span.
• Identity theft complaints rose from 19,778 to 31,675.
• Extortion nearly doubled between 2023 and 2024, from 48,223 to 86,415, then flattened at 89,129.
• Phishing moved the other way, falling from 298,878 in 2023 to 191,561 in 2025.

Investment is the fastest-growing line in the table by a clear margin. Phishing fell over the same period. Reporting behavior appears to have shifted toward the categories carrying real money rather than toward raw volume. The pattern we have documented across regulatory events holds here: enforcement attention follows the loss column, which now points at investment fraud.

Fund Recovery Rates After Crypto Fraud

• The IC3 Recovery Asset Team handled 3,900 incidents through the Financial Fraud Kill Chain in 2025.
• Those incidents represented $1,163,919,846 in attempted theft.
• The team froze $679,013,183, a 58% success rate.
• Domestic actions accounted for 3,574 incidents and $507,042,623 frozen.
• International actions covered 326 incidents and $171,970,560 frozen.
• The team was established in 2018 and works by asking recipient financial institutions to freeze accounts.

Source: FBI Internet Crime Complaint Center 2025 Annual Report (Recovery Asset Team, 2025)

Key finding: Per the FBI’s 2025 report, the Recovery Asset Team froze $679,013,183 of $1,163,919,846 in attempted theft across 3,900 incidents. Domestic actions accounted for $507,042,623 and international actions for $171,970,560, a split showing where the freeze machinery reaches.

Our crypto scam recovery outcomes data tracks how far that picture shifts by payment rail.

Can you recover money from a rug pull?

Published recovery rates are a poor guide here. The team froze 58% of attempted theft value in 2025 by streamlining communications with financial institutions and FBI field offices. A liquidity removal on a decentralized exchange has no recipient bank to contact, so that machinery has no equivalent step to perform.

Rug Pull Enforcement and Criminal Penalties

• A federal jury convicted the SafeMoon chief executive following a three-week trial in May 2025.
• The court sentenced him to 100 months in prison in February 2026.
• The charges were conspiracy to commit securities fraud, wire fraud, and money laundering.
• The sentence included forfeiture of approximately $7.5 million, with restitution to be determined later.
• Prosecutors said the defendant lied to investors and defrauded thousands of victims to buy mansions, sports cars, and custom trucks.

IRS Criminal Investigation announced the sentence with the FBI and Homeland Security Investigations, the agency mix behind enforcement-derived rug pull statistics.

Liquidity removal is not itself the charged act. Deceiving investors about who could reach that liquidity is. Cases turn on what a project said, not on what its contract allowed.

Why are rug pulls illegal?

Prosecutors do not charge a crime called a rug pull. They charge the conduct underneath it, which in the SafeMoon case meant conspiracy to commit securities fraud, wire fraud, and money laundering. The jury also issued a verdict to forfeit two residential properties.

Rug Pulls Inside the Wider Illicit Crypto Economy

• Illicit addresses received at least $154 billion in 2025, the largest annual total on record.
• That represents a 162% increase year over year.
• Stablecoins now account for 84% of all illicit transaction volume.
• The illicit share of attributed volume remains below 1% across the crypto economy.
• Chainalysis stresses the total is a lower-bound estimate based on illicit addresses identified to date.

Source: Chainalysis 2026 Crypto Crime Report (January 2026)

Scale context matters when reading any rug pull estimate. Illicit activity stays a small fraction of measured on-chain volume even at a record dollar total. The harm concentrates rather than spreads. That separates token-launch fraud from the broader Ponzi schemes category, where losses accumulate over months rather than inside an hour.

Is a rug pull legal in the United States?

Federal fraud statutes cover this conduct. No dedicated rug pull offence exists. The SafeMoon case ended with a 100-month prison sentence and an order to forfeit approximately $7.5 million on charges of conspiracy to commit securities fraud, wire fraud, and money laundering.

What the record shows is that prosecutors reach this conduct when they can prove misrepresentation. In SafeMoon, the government secured a conviction following a three-week trial in May 2025 before the sentence was handed down. The legal question is whether the project told holders something untrue about who could reach that liquidity.

Is soft rug pulling legal?

No published enforcement figure isolates soft rug pulls, where a team abandons a project gradually instead of draining a pool at once. The research taxonomy treats abandonment patterns as distinct root causes, and the systematization documents 34 root causes in total. The examined tools reach 25 of them, so nine remain beyond every tool.

That tooling gap mirrors an enforcement gap, because cases reaching sentencing need a provable misstatement and a traceable transfer.

Conclusion

Crypto scams took in at least $14 billion on-chain during 2025, and the FBI recorded $8,648,617,756 in reported investment fraud losses, yet neither total names a single rug pull. The measurement gap runs the length of the topic: detection tools cover 25 of 34 documented root causes, the open datasets that train them cover 7 of 34, and the fund-recovery machinery that produced a 58% freeze rate operates on bank accounts that a liquidity pool does not have.

Two developments will decide whether the next set of rug pull statistics reads differently: detection research now predicts from the first five minutes of trading, and the February 2026 SafeMoon sentence showed that defrauding token investors is prosecutable as securities fraud. Until an agency reports rug pulls as their own category, every total here remains a floor.