AI is transforming crypto scams by letting small teams automate fake identities and conversations, enabling more attacks and higher on-chain revenue while making fraud harder to detect and stop
AI is changing the way crypto scams work. Small groups of scammers can now use automation to impersonate dozens of people at once, run multiple conversations, and target far more victims than before-all without hiring more staff. This shift has made scams both bigger and more sophisticated, as software takes over much of the work that used to require a team of people.
Automation changes the fraud game
Scams like romance fraud, investment impersonation, and business email compromise once took a lot of manual effort. Keeping up a convincing persona and juggling several conversations was time-consuming and required skill. Now, AI has made it possible for a single operator with a good server to do what used to take a whole team. The cost of running a scam has dropped, and it's easier than ever to get started.
Chainalysis found that AI-linked scam operations generated on-chain revenue averaging $3.2 million per scheme-over four times higher than scams without AI involvement.
Chainalysis
Chainalysis data shows the financial impact: scams linked to AI vendors averaged $3.2 million in on-chain revenue, compared to $719,000 for those without AI. Not every AI tool multiplies profits, but automation is clearly attractive to fraudsters. AI can create fake documents, images, voices, and entire digital identities, letting one person manage dozens of targets in different languages. The hardest part-making the scam seem real-can now be handled by software.
Real-world losses and evolving tactics
The FBI's 2025 Internet Crime Complaint Center report logged 22,364 complaints about AI-related scams, with losses totaling about $893.3 million. These cases include fake romantic partners and business impersonators, all using AI to seem more believable. Anthropic's 2025 misuse report described an extortion campaign where its Claude Code model was used to analyze information and draft extortion demands, some for more than $500,000. The company banned the accounts and notified authorities, but the case shows how AI can take over jobs that once needed a human touch in criminal schemes.
Automation hasn't ended scam factories. Amnesty International's 2025 investigation into Cambodian scam compounds found forced labor and trafficking at 53 sites, proving that large-scale operations still exist. FinCEN's review of Southeast Asian scam centers found that traditional criminal groups now work alongside AI-enabled services. Instead of shrinking, the industry is producing more fraud with the same or fewer people. Automation lets operators either cut staff or target more victims-neither outcome helps those being scammed.
In 2025, the FBI Internet Crime Complaint Center received 181,565 complaints involving cryptocurrency, with reported losses of approximately $11.37 billion-a 22% increase from the previous year. This surge highlights the growing intersection of AI-driven fraud and crypto crime, as automation enables criminals to scale up attacks and move funds rapidly across blockchain networks.
FBI IC3 / Cointelegraph
Defenses and the limits of automation
As scammers automate, defenders are trying their own AI tools. O2's Daisy, an AI-powered "grandmother," was built to waste scammers' time by chatting with them for as long as possible. This worked when scammers were human, but loses its effect when both sides use bots. If a scammer's AI spends 40 minutes talking to a defensive bot, no human time is wasted-just electricity and server cycles.
This arms race is pushing attention to the few steps that still need real-world action. Criminals still need accounts to receive stolen funds, payment services to move money, and victims to approve transfers. These are harder to automate. Interpol's 2026 global fraud assessment cited over 1,500 transnational fraud cases with $1.1 billion in reported losses, showing the scale of the problem and the need for payment-blocking measures after fraud is detected. As AI-generated voices and faces become common, users have to do more to verify who they're dealing with, while criminals do less manual work.
Crypto's unique vulnerability
For cryptocurrency users, the risks are even higher. The gap between being convinced and sending money is often just a few clicks, and once funds are sent, they move instantly and can't be recovered. As reported earlier, AI-powered scams are already targeting wallets and browser extensions, swapping out trusted tools for credential-stealing copies. Automation, combined with crypto's speed, makes it easier for scammers to scale up attacks and harder for victims to get their money back.
Chainalysis found that scam operations using AI-linked tools generated on-chain revenue averaging $3.2 million per operation, compared to $719,000 for those without AI. The FBI's 2025 report documented $893.3 million in adjusted losses from AI-related scams, while Interpol's 2026 assessment cited $1.1 billion in losses across 1,500 transnational fraud cases. These numbers show the growing financial impact of AI-driven fraud in crypto.
AI isn't putting scammers out of work-it's making them more efficient and harder to spot. The responsibility for verification and defense is shifting to users, exchanges, and payment providers, while criminals use automation to reach more targets with less effort. As the cost of running a convincing scam drops, the number of attacks is likely to keep rising, and the crypto industry will have to adapt its defenses to keep up.
AI-driven scams highlight a basic trade-off in digital security: as technology makes it cheaper to create convincing fakes, it becomes harder to verify what's real. In crypto, where transactions are fast and final, this creates unique risks for users and platforms. Effective defense will require technical controls, user education, and quick action at the points where real money changes hands. The challenge is not just to keep up with automation, but to make fraud less profitable for attackers.
