The FBI’s 2025 Internet Crime Report shows a materially larger, more sophisticated threat landscape where cryptocurrencies and AI tools were central to large-scale fraud. Total reported losses from cyber-enabled fraud surpassed $21 billion, with more than 1 million complaints filed — a volume and dollar figure that underline both broad exposure and growing attacker sophistication (source: https://www.govtech.com/security/fbi-crypto-ai-scams-drove-billions-in-losses-in-2025).

Headline figures and breakdown

• Total digital-fraud losses: over $21 billion in 2025.
• Crypto-related scams: more than $11 billion, dominated by investment fraud and fraudulent token offerings.
• AI-enabled scams: roughly $893 million, used to produce fake profiles, synthetic videos, and forged documents that accelerate social-engineering successes.
• Business email compromise (BEC): at least $3 billion in losses — a continuing high-dollar vector.
• Romance scams: nearly $930 million.
• Total complaints filed: over 1 million.

How crypto and AI amplify each other

• Crypto enables high-value, near-instant value extraction across borders with limited reversibility. That property makes it attractive for investment fraud, fake exchanges, and illicit cash-outs after account takeovers.
• AI accelerates scale and believability: generative models produce convincing deepfakes, synthetic identities, counterfeit documents, and automated conversational agents. That reduces the time and skill needed to build trust with victims, increasing conversion rates for social-engineering attacks.
• Combined attack chains are common: AI-generated persona + spoofed correspondence (BEC or impersonation) + crypto payment instruction = fast, hard-to-trace loss events.

Market and infrastructure consequences

• Liquidity and custody risk: high-profile crypto scams increase demand for stronger custodial controls and can temporarily reduce market liquidity as exchanges and platforms tighten withdrawal and onboarding policies.
• Regulatory pressure: large aggregate losses draw regulator attention to KYC/AML enforcement, stablecoin oversight, and exchange custody standards — outcomes that affect market structure and product design.
• On-chain tracing vs. off-chain obfuscation: attackers increasingly mix on-chain techniques (tumblers, cross-chain bridges) with off-chain services (OTC desks, peer-to-peer trades) to monetize stolen assets faster, pressuring chain-analysis firms and compliance teams to adapt.

Operational mechanics attackers use

• Investment fraud: fake token launches, pump-and-dump coordination, fraudulent “guaranteed return” schemes paired with social-media amplification.
• BEC and impersonation: credential compromise used to authorize wire or crypto transfers; AI-generated voices/videos used to validate requests.
• Romance and extortion: long-conversion social-engineering that culminates in requested crypto transfers or forced liquidation of held assets.
• Synthetic identity chains: AI-generated identities scaled via automated account creation to bootstrap credibility across multiple platforms.

Defensive measures that materially reduce exposure

• Wallet hygiene and custody segregation: strict separation of hot/cold keys, whitelisting withdrawal addresses, multisig for outbound transactions, and time-locked withdrawals to allow human review on large transfers.
• Transaction-level controls: per-address and per-transaction limits, anomaly detection that flags new counterparty patterns, and mandatory hold periods for first-time off-chain recipients.
• Identity and communication verification: multi-channel validation for fund movements, hardened procedures for any request tied to leadership or finance teams, and mandatory secondary confirmation for changes to payment instructions.
• AI-aware controls: use synthetic-identity detection, deepfake detection tools for high-risk onboarding, and enhance behavioral analytics to detect automated account activity.
• User and employee training: periodic phishing and deepfake-awareness exercises; simulated compromises to harden response workflows.
• Reporting and partnering: timely reporting of incidents to law enforcement/compliance bodies and coordination with exchanges and chain-analysis firms to freeze or trace assets quickly.

Token-design note (relevant to liquidity management)
Some token economic designs aim to reduce immediate sell pressure and create predictable liquidity behavior by using fixed-price entry, short predefined holding cycles, and staged unlock mechanisms. For projects where predictable timing matters, mechanisms that reward early participation while staggering unlocks can lower the likelihood of sudden dumps and make abnormal flows easier to detect by compliance systems (an example is a model that uses fixed entry pricing and programmed unlocks to encourage disciplined timing).

Tactics for incident response and recovery

• Immediate containment: freeze custodial wallets where possible, engage exchanges and on-chain analytics firms to trace flows, and escalate to law enforcement.
• Preserve evidence: capture message threads, recorded calls, and digital artifacts — AI-generated content is often identifiable by metadata and synthesis artifacts useful to investigators.
• Remediation playbooks: predefine roles, legal coordination, customer notification procedures, and steps for asset recovery or civil action.

Regimes that will matter going forward

• Stronger AML/KYC practices at on- and off-ramps, combined with faster cross-border information sharing, will increase the cost and reduce the speed of illicit cash-outs.
• Platform liability and consumer-protection rules will push custodians and marketplaces to bake in safeguards such as mandatory waiting periods, enhanced transaction screening, and clearer dispute-resolution pathways.

Source reference: https://www.govtech.com/security/fbi-crypto-ai-scams-drove-billions-in-losses-in-2025