The FBI’s latest Internet Crime Report quantifies what market operators and compliance teams have been warning about: emerging technologies are multiplying attack vectors and amplifying losses. In 2025 the agency attributes more than $21 billion in reported losses to internet-enabled fraud, with cryptocurrency and AI-enabled schemes driving a disproportionate share of the damage.

Digital-asset channels dominated the tab. Crypto-related fraud exceeded $11 billion, with investment scams accounting for roughly $8.6 billion of that total. The report notes that nearly three‑quarters of all cyber-fraud incidents involved cryptocurrency at some stage—either as the requested payment method or as the laundering rail. That degree of prevalence has direct implications for market confidence and for intermediaries that onboard retail flows.

AI-enabled deception is emerging as a high-velocity enhancer of traditional fraud methods. The FBI attributes about $893 million in losses to scams using AI tools and received roughly 22,000 complaints tied to that class of attacks. Common tactics include generating realistic fake messages, cloning voices for phone scams, and creating synthetic profiles that bypass basic social checks—techniques that scale social‑engineering success rates and compress the time between contact and payment.

Other categories remained material. Business email compromise (BEC) caused at least $3 billion in losses, with high-profile cases such as a school district defrauded of $4.92 million. Government-impersonation scams generated more than 32,000 complaints. Tech‑support scams and extortion schemes continued to extract significant sums, often leveraging the same AI tools and identity-spoofing techniques cited above.

Mechanics and threat evolution

• Why crypto is attractive to fraudsters: speed, finality, and pseudonymous settlement rails reduce the window for reversal and complicate recovery. Criminals increasingly combine on-chain transfer techniques—mixers, privacy coins, cross‑chain bridges and rapid decentralised exchange swaps—to obscure provenance.
• Why AI matters: automated content generation and voice cloning lower the cost-per-target and increase conversion. AI lets attackers mimic trusted parties with higher believability and faster iteration, enabling campaign scaling that manual social engineering cannot match.
• Convergence risk: attackers blend channels—BEC messages that direct victims to pay in crypto, AI-generated impersonations that win trust, and laundering pipelines that rapidly move proceeds—making single-point defenses less effective.

Implications for markets and defenders

• Recovery and traceability: On‑chain analytics have improved recovery and case attribution, but speed is decisive; the faster funds move through layered liquidity, the lower the recovery probability. Exchanges and on‑ramps remain critical choke points for both prevention and remediation.
• Compliance posture: Existing KYC/AML controls need to be married with behavioral and transaction‑pattern analytics that spot investment-scam indicators (high-volume retail inflows to new token contracts, rapid outbound routing through mixers, repeated small withdrawals into darknet wallets).
• Organizational hygiene: BEC and impersonation losses underscore gaps in internal payment controls and verification processes. Multi-step approvals, out-of-band confirmations for wire transfers, and continuous staff training materially reduce success rates for these attacks.

Actionable priorities the FBI highlights

• Report quickly: Rapid reporting increases options for tracing and freezing funds. The agency emphasizes immediate notification to law enforcement and financial partners when a compromise is suspected.
• Public awareness and training: Scaled, recurring training for employees and the public on AI-enabled deception, voice-cloning risks, and red flags for crypto investment pitches.
• Private–public collaboration: Faster intelligence sharing between exchanges, chain‑analytics firms, and law enforcement to disrupt laundering flows and identify actor infrastructure.

For market participants and custodians, the economic lesson is straightforward: as illicit actors adopt faster, more convincing tools, defenders must compress detection and verification timeframes and build layered, protocol‑aware controls. Source: https://www.govtech.com/security/fbi-crypto-ai-scams-drove-billions-in-losses-in-2025