TRM Labs: AI Is Supercharging Crypto Fraud

TRM Labs is warning that crypto fraud is entering a more dangerous phase — one driven by AI, automation, and speed.

In a new report, the blockchain intelligence firm says artificial intelligence is turning crypto-enabled fraud from a labor-heavy operation into a machine-scaled system, where scammers can launch, test, and adapt campaigns much faster than before. TRM says that shift is already showing up in the numbers.

AI is changing the economics of fraud

TRM’s main point is not just that scams are getting smarter — it’s that they’re becoming easier to run at scale.

According to the report, AI tools are helping fraud actors automate phishing, impersonation, and identity-based deception, while also improving personalization and response speed. In practical terms, that means more convincing scams, faster iteration, and lower operational cost for criminal networks.

That matters because fraud used to require more human effort at every step. TRM’s framing is that AI is removing that bottleneck.

TRM’s numbers: illicit volume up, scam activity still huge

TRM says illicit crypto activity reached a record $158 billion in 2025, up nearly 145% year over year. Within that, the firm estimates scam-related activity at around $30 billion, while cautioning that underreporting likely means the real number is higher.

The report also says TRM observed an approximately 500% increase in AI-enabled scam activity over the past year, which it presents as evidence that generative AI is now being integrated into fraud operations at meaningful scale.

What “AI-enabled fraud” means

TRM defines AI-enabled fraud as scams and financial crimes that use AI to automate, personalize, and scale deception.

The report’s framing is useful because it avoids the usual hype. This isn’t a claim that AI “creates” crypto crime. It’s a claim that AI makes existing scam models harder to detect and easier to deploy — especially in crypto, where payments are global, fast, and hard to reverse.

Why this is a bigger problem for the industry now

TRM ties the shift to the combination of three things: generative AI, programmable money rails, and global crypto liquidity.

That combination gives bad actors a dangerous stack: AI handles social engineering and fake identities, while crypto handles fast movement of funds across borders. The result is a fraud environment that can scale faster than many compliance and investigative teams are used to.

For exchanges, wallets, and compliance teams, the takeaway is pretty direct: scam detection and incident response now need to assume adversaries are using AI by default.

Why it matters for crypto

• TRM is signaling that AI fraud is no longer a niche risk — it is becoming part of the core threat model for crypto platforms.
• The reported 500% rise in AI-enabled scam activity suggests fraud defenses built for manual scam operations may now be too slow.
• TRM’s $158 billion illicit volume estimate (with scams at ~$30 billion) reinforces that fraud remains one of crypto’s biggest trust and policy challenges.
• The report strengthens the case for better monitoring, identity checks, and faster cross-platform coordination as stablecoins and crypto payments scale further. (This is an industry inference based on TRM’s findings.)

What to watch next

• Whether TRM and other analytics firms publish more granular breakdowns of which scam types are growing fastest under AI-enabled models.
• How regulators and law enforcement respond as AI-driven fraud becomes a bigger part of crypto policy discussions. (Inference based on the scale described in the report.)
• Whether exchanges and wallet providers start rolling out more visible anti-impersonation and anti-social-engineering safeguards in user flows. (Inference based on the risks TRM describes.)
• New data on underreporting, since TRM explicitly notes scam losses are likely materially higher than reported totals.

What we don’t know yet

• TRM’s public summary does not provide a full methodology breakdown in the excerpt for how the 500% AI-enabled scam increase was measured.
• The report excerpt highlights the scale of the problem, but not a detailed split of losses by chain, region, or scam subtype.

Source: TRM Labs blog