Cryptocurrency exchange Bybit says its new AI-driven on-chain monitoring and risk framework successfully intercepted and recovered about $300 million of flagged suspicious withdrawals tied to scams and fraudulent activity during the fourth quarter of 2025. The initiative is part of the platform’s broader 2025 Security Initiative, aimed at detecting and stopping fraud before funds are lost.

According to Bybit, its systems flagged roughly $500 million in high-risk withdrawal attempts, of which approximately $300 million was prevented from leaving user accounts. The exchange reported that more than 4,000 users were directly protected, with its AI algorithms identifying hundreds of high-risk addresses based on on-chain patterns and behavior. In addition to blocking scam transactions, Bybit said the monitoring framework also detected tens of thousands of credential stuffing and impersonation attack attempts throughout 2025.

Bybit’s multi-tier risk architecture integrates artificial intelligence models trained to recognise abnormal wallet activity, suspicious transaction flows, and scam-related indicators in real time. When the system detects a potentially fraudulent withdrawal, it triggers alerts, automated cooling-off periods, or blocks the transaction until risk teams can investigate. The framework is reinforced by data from blockchain analytics firms, which help label and track known scam addresses and addresses linked to widespread fraud schemes.

Industry analysts see this development as part of a broader trend in crypto security, where advanced monitoring technologies are increasingly used to protect retail and institutional holders from scams. In 2025 overall, global crypto fraud losses including phishing, impersonation schemes, and other social engineering tactics were estimated in the billions, emphasising the value of proactive defence systems. Bybit’s performance metrics suggest that automated detection and intervention tools may help reduce the impact of fraud in on-chain environments and improve platform reputation and user confidence.

The success of Bybit’s system reflects how exchanges are adapting to increasingly sophisticated fraudulent techniques, including those enabled by automated scripts and social engineering attacks. While these tools do not eliminate all risk, they represent an evolving security arms race between defenders and emerging threats within the digital asset ecosystem. For market participants, such developments may bolster sentiment around platform safety and highlight the importance of robust compliance and monitoring infrastructure in maintaining trust in centralized trading venues.