AI-Powered Crypto Phishing Changes the Economics of Attacks

AI has altered the unit economics of crypto phishing. Attackers can now produce convincing lures at near-zero marginal cost, personalize them at scale, and convert more targets. The clearest signals are rising average payments, explosive growth in impersonation scams, and the tight coupling of AI with phishing-as-a-service kits.

Chainalysis estimates at least $14 billion flowed to on-chain scam addresses in 2025 and projects total scam and fraud losses could reach $17 billion. The average scam payment jumped 253% year over year from $782 in 2024 to $2,764 in 2025. Impersonation scams grew about 1,400% year over year, and Chainalysis found AI-enabled scams were roughly 4.5 times more profitable than traditional scams (Chainalysis).

Broader cybercrime telemetry points the same way. The FBI’s IC3 recorded 1,008,597 total complaints in 2025 and flagged 181,565 cryptocurrency-related complaints with reported losses of $11.366 billion. It also tracked AI-related complaints for the first time, logging 22,364 complaints and $893.346 million in reported losses (FBI IC3 2025 Annual Report). On the distribution side, Barracuda’s analysis of more than 3.1 billion emails found one in three email messages were malicious or unwanted, 48% of malicious email was phishing, and 90% of high-volume phishing campaigns used phishing-as-a-service. Barracuda reports adversaries are pairing AI-driven social engineering with PhaaS to scale targeted credential harvesting (Barracuda).

… Continue reading the full article at the original source below.