Bybit Blocks $300 Million in Fraud Funds in 2025

Bybit announced the results of its 2025 security plan on February 27. The crypto exchange revealed it intercepted roughly $300 million in funds linked to fraud, impersonation and fraudulent activities last year via a new AI-powered risk control framework. Its "Three-Tier Withdrawal Anti-Fraud Defense System" is built on a dynamic risk grading mechanism, with three levels: - **Level 1 (Low Risk):** Uses big data heuristics to spot abnormal patterns (e.g., large transfers to new addresses), automatically sends a questionnaire for review, and pre-blacklists high-risk addresses. - **Level 2 (Medium Risk):** Triggers a real-time alert during withdrawals if an account is tied to leaked databases or suspicious addresses, prompting users to review the transaction to block social engineering scams. - **Level 3 (High Risk):** Implements real-time withdrawal interception for confirmed fraudulent addresses, plus a mandatory 1-hour cooling-off period. ### Key 2025 Q4 Data - Of $500 million in suspicious withdrawals, $300 million was successfully intercepted and recovered, involving over 4,000 users. - The AI algorithm flagged 350 high-risk investment scam addresses, protecting 8,000 users. - Over 3 million account takeover attacks were blocked throughout the year. - 350 addresses were auto-flagged, 600 manually flagged, avoiding potential losses of roughly $1 million.