Live
Loading prices…
CoinHub Today · coinhubtoday.com
CoinHub Today · Security

AI-Powered Phishing Is Eating Crypto: $311M Drained in January Alone

Deepfaked support agents, cloned executive voices, and wallet-drainer kits rented by the hour. Generative AI has made impersonation scams a 24/7 industrial pipeline — and the retail side of crypto is paying the price.

SecurityAIPhishingCoinHub Today Research DeskApril 10, 20267 min read

Phishing used to be the unsexy tail of crypto crime — slow, manual, and generally easy to spot. That is no longer true. In January 2026 alone, phishing attacks drained more than $311 million from crypto users, according to Chainalysis — more than six times the monthly average two years ago, and a record for the category. The driver is generative AI, and the evolution has been startlingly fast.

$311M
Phishing losses in January 2026 alone
1,400%
Impersonation scams YoY increase
$30M
Monthly wallet-drainer losses (ScamSniffer)
90 sec
To produce deepfake video call
TacticHow it worksTypical loss
Deepfaked exec video callsAI-cloned CTO requests emergency transfer$500K-$5M per incident
AI voice-cloned support agentsFake call from 'exchange security' extracts codes$10K-$200K retail
Malicious AI-generated dApp frontendsIndexable clone sites with working UX$5K-$100K per victim
Browser-extension wallet drainersRented kits, AI-tuned approval prompts~$30M/month industry-wide
Telegram / Discord persona clonesAI-scraped identity + voice clone + deepfake video$20K-$1M per incident

The Deepfake Threshold Has Been Crossed

In late February, a mid-sized crypto market-making firm lost $4.8 million when an engineer joined what he believed was a scheduled video call with his CTO and CFO. Both executives on the call were AI-generated, running real-time lip-sync against voice clones trained on podcast appearances and conference talks. The engineer signed the transfer, hung up, and learned about the compromise an hour later when the real CTO walked into the office.

"The attackers have crossed a line that matters. Deepfake quality is now inside the uncanny-valley envelope. In a normal work context, it is genuinely indistinguishable from reality. You cannot train humans out of this. You have to redesign the trust architecture."

— Taylor Monahan, MetaMask

The Industrial Pipeline

Wallet-drainer kits — rented by the hour on underground forums and increasingly integrated with LLM-driven social-engineering scripts — account for roughly $30 million a month in industry-wide losses per ScamSniffer. A typical kit now ships with AI-generated phishing email templates, a malicious frontend cloned from a popular dApp, and a prompt library tuned to produce approval requests that users will habitually sign.

The Defensive Response

Coinbase, Kraken and Bitstamp all rolled out out-of-band authentication for high-value support interactions in Q1. Several wallet providers have deprecated email as a 2FA channel. TRM Labs launched a product that clusters wallet-drainer operators by on-chain laundering signatures, flagging roughly 1,800 addresses tied to active campaigns. A global raid backed by TRM Labs froze $12 million and identified more than 20,000 victims of a single cluster.

"The supply of potential targets is the entire internet. We are running against a conveyor belt."

— Ari Redbord, TRM Labs Global Head of Policy

Crypto's pseudonymous, irreversible, self-custodial model makes the human-engineering layer far more dangerous than it is in traditional finance. AI has turned that attack surface into an industrial-grade pipeline. And the industry's defenses, however improved, have not yet caught up.

The Structural Problem
The industry has spent years arguing that most retail losses have nothing to do with smart-contract bugs. In 2026, that defense stops working. AI has made impersonation scams faster to deploy, harder to detect, and cheaper to run than at any previous point in history.
Sponsored · Web3Firewall
Pre-Signature Intelligence for Crypto Operations

The attacks described in this article exploit gaps that pre-signature transaction monitoring is built to close. Web3Firewall evaluates 100+ risk signals before a transaction reaches the blockchain — enforcing policy controls at the only moment intervention is actually possible.

Reporting note: Draws on public disclosures from Chainalysis, TRM Labs, Elliptic, CertiK, Halborn and affected protocols. Editorial commentary; figures subject to revision as investigations continue.

Stay ahead with CoinHub Today

Real crypto news, market data, and analysis — free to your inbox every weekday at 7am.

No spam. Unsubscribe anytime. Sent to admin@coinhubtoday.com

CoinHub Today is an independent media organisation and does not provide investment, financial, or legal advice. All content is for educational purposes only. Cryptocurrency investments involve substantial risk. Past performance is not indicative of future results. Always consult a qualified financial adviser before investing.