AI-Driven Crypto Scams Spike in 2025–26

Reports show AI-enabled crypto scams jumped 456% between May 2024 and April 2025, and Chainalysis data indicate roughly 60% of deposits to scam wallets now flow to schemes that use generative AI. The figures reflect wider adoption of large language models, deepfake video and audio, voice cloning and automation by fraud operators in 2025.

Security firms report that AI tools let a single operator produce thousands of personalized phishing messages, fake sites and impersonations in minutes. AI-generated content often avoids traditional red flags such as poor grammar and reused domains, and can be customized with public information about targets. Newer threats include prompt-injection attacks against agentic browsers and wallet assistants that can read pages or act for a user; hidden instructions in a webpage or image can cause those assistants to reveal data or trigger transactions.

Cryptocurrency is a frequent target because transactions are fast and often irreversible, markets operate 24/7 and many users trade outside standard consumer-protection frameworks. Scammers use urgent offers like exclusive airdrops, celebrity endorsements or promised returns to push victims to act quickly. Chainalysis data show AI-backed methods are responsible for a growing share of illicit inflows to scam wallets.

Common schemes now mix deepfakes, large language models and automation. Deepfake audio and video have been used to impersonate executives and celebrities to instruct victims to transfer funds. One case involved a finance employee in Hong Kong who transferred $25 million after appearing on a video call with what looked like company executives; investigators later determined the faces and voices were generated by AI. Tools such as WormGPT and FraudGPT have been cited as enabling operators to scale convincing attacks.

Other methods include AI-generated phishing that produces highly personalized messages and fraudulent websites, fake trading platforms or bots that claim automated profits, and voice cloning for live phone calls. Long-running “pig-butchering” cons that build trust over weeks or months have also incorporated AI to sustain believable personas. Chainalysis reported that AI-assisted pig-butchering scams generated billions globally in 2024.

Fraud groups also use AI to create fake identity documents and selfies to bypass know-your-customer checks and open mule accounts. Large social botnets imitate real accounts on platforms such as X to push wallet-draining links and false giveaways.

Security guidance for users includes enabling two-factor authentication or passkeys on email, exchanges and wallets, keeping seed phrases offline and private, using hardware wallets to store private keys, accessing platform support only through official websites or apps, and avoiding links or wallet approvals requested in unsolicited messages. Analysts note detection systems are being updated to address more sophisticated content, while basic precautions remain a primary defense for many users.