Binance Says AI Defenses Blocked $10.5 Billion in Crypto Fraud Over 15 Months
Binance, the world's largest crypto exchange, says it has deployed more than 100 AI models to counter a surge in AI-powered scams.
By Decrypt AgentEdited by Andrew HaywardMay 11, 2026May 11, 20262 min read
In brief
- Crypto scams hit $17 billion in 2025, as AI makes attacks cheaper and more convincing than ever.
- Binance says 100+ AI models helped block $10.5 billion in fraud and intercepted 22.9 million scam attempts over 15 months.
- Beyond prevention, Binance said it recovered $12.8 million and helped authorities confiscate $131 million in illicit funds in 2025.
Binance, the world's largest cryptocurrency exchange by trading volume, said Monday that its AI-driven security systems prevented more than $10.5 billion in user losses between the start of 2025 through the first quarter of 2026, as fraudsters increasingly weaponize the same technology to launch attacks at unprecedented scale and speed.
The exchange said its systems intercepted 22.9 million scam and phishing attempts in the first quarter of 2026 alone, safeguarding approximately $1.98 billion in user funds.
The disclosures come as AI is dramatically lowering the cost and complexity of crypto fraud. According to Binance's own research, smart contract exploits now cost bad actors as little as $1.22 per contract—down 22% month-over-month—while advanced AI models achieve a 72.2% success rate in attack scenarios. The company said crypto-related fraud reached $17 billion in 2025, a 30% increase from the prior year.
The exchange said 76% of AI-driven scams now fall within the highest tier for both scale and severity, with attackers deploying deepfakes, voice cloning, phishing bots, and impersonation schemes across messaging platforms to exploit user trust.
To counter those threats, Binance said it had deployed more than 24 AI initiatives and over 100 models by late 2025. The company said AI-driven systems now power 57% of its fraud controls, contributing to a 60-70% reduction in card fraud rates compared to industry benchmarks.
The company also highlighted a new product, Binance AI Pro, which is designed to contain risk at the architecture level. Under that framework, funds managed by AI agents are segregated from main user accounts, with permissions limited to trading only and no withdrawal access. The company said roughly 12% of third-party tools submitted to its marketplace have been flagged as potentially risky.
Recovery efforts are also expanding. Binance said it helped recover $12.8 million across 48,000 cases in 2025—a 41% year-over-year increase—and assisted authorities in confiscating $131 million in illicit funds while processing more than 71,000 law enforcement requests.
“AI is reshaping both sides of the security equation. It’s making attacks more scalable, more convincing, and harder to detect, while also enabling a new generation of defenses that are faster, smarter, and more adaptive,” the firm wrote. “To close the gap between exploitation and detection, security must evolve at the same pace, embedded across systems, processes, and user behavior rather than treated as a separate layer.”