1. TRM Labs says that losses from fraud using "deepfake videos" reported in 2026 have already exceeded the total for all of 2025 by 263%.
2. Identity impersonation supported by artificial intelligence targets account recovery decisions, signatures, and payments, bypassing the protections that secure accounts, keys, and smart contracts.
3. The remaining challenge is to verify the true human intent before authorized transfers become irreversible—especially after changes to identity data or withdrawal details.
Maturing AI scams alongside identity impersonation expansion
According to blockchain intelligence firm TRM Labs, losses reported from scams involving deepfakes in 2026 exceeded the total of the prior year by 263%, highlighting a growing security problem in the cryptocurrency world, where attackers are increasingly turning to manipulating authorized users rather than hacking blockchain code.
The company’s new index, the "AI-in-Crime Adoption Index," classifies scams as the only category of cryptocurrency-related crimes in which AI use has reached a "mature" level. TRM said that reports about fraudsters using AI—including deepfakes, chatbots, and AI-powered traps—have increased by about 13 times since 2022.
This shift exposes a gap that traditional smart-contract security does not address: the trading account might be properly authenticated, the hardware wallet might sign correctly, and the smart contract might execute exactly as programmed—yet funds can still reach an attacker if a deepfake video succeeds in persuading the person controlling these systems to approve the transaction.
This shifts the greater security burden to the moment just before the declaration, when a trading platform decides whether a request to restore an account is genuine, or when one of the authorized signatories agrees—by signing—to a transfer, or when a person accepts a payment instruction from someone they believe they know.
Measuring scam maturity
The TRM index measures how widely AI is spread across different types of crimes, and how broadly it’s used across stages such as targeting and deception, as well as how advanced the tools used are. The company said that its broadest dataset—covering all fraud reports that mention AI—has grown by about 25 times since 2022 (this includes cases where victims used consumer AI tools while investigating suspected fraud). The narrower 13x increase is limited to reports in which the fraudsters themselves used AI.
Supporting data from other sources
Other data point in the same direction:
- Chainalysis: flows to identity-impersonation scams rose by more than 1,400% year over year, and it found that scams with blockchain-visible links to AI service providers generated, on average, 4.5 times higher revenue than those without such links.
- The Federal Bureau of Investigation (FBI), in its 2025 report on internet crimes: recorded 22,364 complaints related to the description of "artificial intelligence," with reported losses of $893.35 million. Separately, losses from complaints related to cryptocurrency totaled $11.37 billion.
The breach increasingly occurs before signing
The problem becomes more severe in the cryptocurrency world because transactions are difficult to reverse once they’re authorized. A separate review by TRM of breaches in the first half of the year found that smart contract vulnerabilities remain common, but the biggest losses were concentrated in infrastructure breaches and operational compromises—which may involve stolen credentials, private keys, or other forms of access that allow the attacker to issue instructions that the blockchain network treats as legitimate.
These deepfakes expand the problem by helping attackers obtain the victim’s "cooperation" instead of simply stealing access. On one trading platform, an attacker can impersonate a customer during account recovery, change authentication factors, and add a new withdrawal destination—each subsequent step can seem correct because the attacker has already compromised the identity verification decision that controls access.
FinCEN’s network has warned financial institutions about monitoring mismatches in identity data, suspicious device or location changes, third-party webcam tools, resistance to multi-factor authentication, and rapid transactions following account changes.
Corporate vaults are also exposed
An artificial voice or video of an executive can pressure an employee to approve a transfer, change one of the signatories, or add a new payment address. Hard wallets can confirm that the correct private key—one that signed the transaction—is being used, but they cannot determine whether the person controlling that key has been deceived.
Multi-person approval, pre-configured confirmation channels, and delays before newly added withdrawal addresses are activated can move the decisive decision out of the attacker-controlled communication channel.
The FBI also warned that North Korean IT workers used fake identities, manipulated videos, AI tools, and remote-access infrastructure to obtain positions that granted them privileged access to corporate systems and cryptocurrency.
For individuals
For individual cryptocurrency holders, the attack can be far simpler: a convincing video call, a voice message, or a fake profile may convince the victim to make the payment themselves— and in this case, blockchain monitoring doesn’t begin until after the crucial security failure has already occurred.
Conclusion
Blockchain monitoring tools remain useful for spotting suspicious flows, tracking stolen assets, and supporting freeze operations where central intermediaries can intervene. But they are less effective at stopping a transaction that appears legitimate because the victim—or the authorized signer—agreed to it voluntarily. Cryptocurrency companies still need smart-contract audits, protection of private keys, wallet simulation, and transaction monitoring—but with the increasing effectiveness of AI identity impersonation, the most important control tool may increasingly be one that challenges and verifies the identity of the instruction issuer before signing an irreversible transaction.