AI is becoming a critical defence for crypto exchanges as fraud losses surge and criminals use the same technology to make scams harder to detect.
Illicit crypto wallets received $158 billion in 2025, up 145% from $64.5 billion in 2024, reversing the decline seen since 2021, according to TRM Labs' 2026 Crypto Crime Report. Impersonation scams alone rose 1,400% from a year earlier.
The shift is changing the nature of crypto fraud. Criminals are moving from mass, low-value scams towards fewer, larger targets, using generative AI to create convincing identities, messages and documents. Exchanges are responding with AI systems that monitor transactions, account behaviour and identity checks in real time.
TRM Labs and Chainalysis both link the increase in scam activity to generative AI. Scam operations with visible on-chain links to AI vendors generated 4.5 times more revenue per operation than those without such links.
Deepfakes can now create fake customer-support agents, government notices using officials' likenesses and convincing impersonations of trusted insiders at a scale that manual checks cannot match.
Losses Mount
The wider fraud picture shows the same trend.
The FBI's April 2026 report attributed $893.35 million in adjusted losses to US complaints that referenced AI in 2025. The complaints covered 22,000 filings.
Investigators believe the figure understates the actual losses because victims of AI-driven phishing and voice-cloning scams often do not identify the technology used to defraud them.
Europol's 2025 Internet Organised Crime Threat Assessment projected that about 8 million deepfakes would be shared online during the year, up from about 500,000 in 2023.
Gartner expects nearly one-third of enterprises in 2026 to stop trusting identity-verification systems that rely only on facial biometrics because of the growing ability of deepfakes to defeat them.
Prevention First
The nature of blockchain settlement has made prevention more important for crypto platforms.
Transactions can be close to irreversible once funds move. That reduces the value of recovering money after a fraudulent transfer and puts greater pressure on exchanges to identify suspicious activity before settlement.
The range of fraud methods is also expanding. Phishing, fabricated investment schemes, identity theft and account takeovers are becoming more sophisticated as platforms strengthen their defences.
Wallet-drainer kits, for example, can direct victims to fake NFT marketplaces and DeFi platforms to extract funds.
Digital assets have also moved deeper into mainstream finance. Retail investors, institutions and companies are participating in greater numbers, expanding the potential pool of targets for fraudsters.
Fraud groups use AI to write convincing scripts, clone voices and generate fake identity documents. Exchanges are using AI to identify those same tactics.
Neither side has gained a decisive advantage. But exchanges can now monitor activity across markets and block suspicious behaviour at a scale that manual systems could not support.
Real-Time Checks
Alan Xin, head of blockchain risk control at Bybit, said transaction volumes have made manual oversight inadequate.
"Exchanges process millions of transactions daily across markets and blockchains, a volume that outgrew manual compliance years ago. That gap between transaction volume and oversight capacity is precisely where AI has found its most immediate use case," Xin said.
AI models can continuously examine transaction flows, login activity, device fingerprints and account behaviour. They can flag large withdrawals, unfamiliar login locations and trading patterns that differ from a user's history.
Platforms can then assign a risk score to the activity and respond according to the level of risk. They may pause a transaction or require additional verification before allowing it to proceed.
The aim, Xin said, "is to intervene early enough that the damage never occurs in the first place."
Phishing Defence
Phishing is another area where exchanges are turning to AI.
Fraudsters regularly create fake websites, spoofed emails and deceptive messages designed to steal credentials or seed phrases. Static block lists struggle to keep pace because scam websites and messages can change frequently.
AI systems can instead identify patterns associated with fraudulent activity rather than relying only on known malicious URLs. They can also retrain as fraud tactics change.
That allows platforms to intervene before credentials are exposed.
Identity Checks
Identity verification is pushing AI beyond fraud detection and into compliance infrastructure.
Know Your Customer checks remain a core requirement, but manual document reviews can miss sophisticated forgeries. AI-powered systems can process documents faster and, according to Xin, with greater accuracy, while reducing risk and speeding up onboarding.
Anti-money laundering is another major use case.
Illicit funds rarely remain in one wallet. Criminals can move money through multiple addresses and exchanges to obscure its source, operating at blockchain speed rather than traditional banking speed.
AI can track those movements even when individual transactions do not appear suspicious on their own. That reduces the time required to reconstruct a flow of funds and moves the process closer to real-time monitoring.
Blocking Scams
Chainalysis' Alterya platform monitors transactions and can block transfers to active scam accounts before funds leave a customer's wallet.
It is part of a growing group of blockchain analytics tools designed to detect fraud in real time.
SEON's 2026 industry survey covering payments, fintech, banking and gaming found that 98% of organisations already use AI in fraud and anti-money laundering workflows.
Tamas Kadar, chief executive of SEON, argues that the question is no longer whether AI can work in fraud prevention.
The scale of the threat continues to rise. Nearly 150 hacks in 2025 resulted in combined losses of $2.87 billion.
AI is making deception cheaper for criminals while making it more convincing. The World Economic Forum's Global Cybersecurity Outlook 2026 identifies AI as the dominant force reshaping cyber risk.
The response from fraud teams is not simply greater automation. Crypto exchanges are moving towards screening transactions before they settle, using AI to assess risk continuously rather than investigating suspicious activity after funds have already moved.
For an industry built around fast, often irreversible settlement, that shift could become a core requirement for its next phase of institutional growth.
Disclaimer: The views expressed in this article are solely those of the author and do not necessarily reflect the opinion of NDTV Profit or its affiliates. Readers are advised to conduct their own research or consult a qualified professional before making any investment or business decisions. NDTV Profit does not guarantee the accuracy, completeness, or reliability of the information presented in this article.
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.
