Binance says it stopped $10.53 billion in user losses across 15 months using AI-driven fraud controls, the exchange filing the numbers on Monday. The figure covers activity from Q1 2025 through Q1 2026. Over half the platform’s fraud defences now run on automated decisioning rather than manual review.
The exchange says it protected 5.4 million users from scams and blacklisted 36,000 malicious addresses. In Q1 2026 alone, the system intercepted 22.9 million phishing and scam attempts, saving $1.98 billion in user funds according to the post.
Binance AI fraud controls scale as attack sophistication rises
AI-powered scams are accelerating. That’s the opening line in the exchange’s update. The barrier to entry has dropped. What used to require technical skill can now be executed at scale for almost nothing, the exchange says.
The FBI reported $11 billion in US crypto losses to fraud in April. Impersonation of officials or platforms remains a primary attack vector. Voice cloning, deepfakes, phishing bots. The threat actors are using the same technology the defences are deploying.
| Metric | Period | Result |
|---|---|---|
| User losses prevented | Q1 2025 to Q1 2026 | $10.53 billion |
| Scam attempts blocked | Q1 2026 | 22.9 million |
| Malicious addresses blacklisted | Q1 2025 to Q1 2026 | 36,000 |
| Users protected | Q1 2025 to Q1 2026 | 5.4 million |
| Fraud controls using AI | Current | 57% |
Binance rolled out more than 24 AI-driven initiatives and over 100 models during the period. Computer vision now scans for fake payment proofs. Real-time language analysis flags scam patterns. On the identity verification side, the technology counters deepfakes and synthetic identities, the exchange says.
Card fraud rates down 60% to 70% against industry levels
AI-driven decisioning now powers 57% of fraud controls at the exchange. Card fraud rates have dropped 60% to 70% compared to industry benchmarks, according to the update.
The crypto sector has seen highly organised threat actors adopt AI to create more sophisticated attacks. Social engineering is scaling at an unprecedented level, the exchange says. Phishing bots, fake platforms, voice cloning, impersonation across messaging applications. The tactics exploit trust and urgency.
The exchange framed the shift as a necessary response to a collapsing barrier to entry for attackers. What once required expertise is now commoditised. The fraud controls are scaling in parallel.
The read
The numbers are large. $10.53 billion is not a rounding error. Whether the 57% AI deployment figure represents genuine automation or just model-assisted flagging is not detailed. The card fraud comparison lacks a named benchmark, so the 60% to 70% claim is difficult to verify.
The trend is clear. The arms race between fraud tooling and defence tooling is accelerating. Both sides are using the same technology stack. The exchange is publishing the numbers to signal capability. Whether the industry adopts similar controls or continues to see losses depends on how quickly the laggards catch up.
Next milestone: watch for regulatory frameworks around AI deployment in fraud detection. The Financial Conduct Authority and US Securities and Exchange Commission are likely to weigh in on transparency requirements for automated decisioning in financial services. The crypto sector does not operate in a vacuum.
This article is for information purposes only and does not constitute investment advice. Readers should not act on any information contained here without first consulting an authorised financial adviser. Past performance is not a reliable indicator of future results.
