Bybit, a major cryptocurrency exchange, has reported that its newly implemented AI-powered fraud detection system prevented over $300 million in potentially fraudulent withdrawals during the final quarter of 2025. The system is designed to identify and halt malicious transactions before digital assets can leave the platform, marking a significant shift toward preemptive security measures in the crypto industry.
The exchange detailed that its risk monitoring framework flagged approximately $500 million in withdrawal requests over the three-month period. According to the company, more than 4,000 users were protected after receiving real-time risk alerts, with many choosing to cancel their transactions voluntarily upon seeing the warnings. This proactive approach means funds often remained securely in user accounts without the need for complex recovery efforts.
David Zong, Bybit's head of group risk control, emphasized that a substantial portion of the $300 million figure represents withdrawals users canceled themselves. This highlights the effectiveness of warning systems in preventing losses before they occur, moving beyond the traditional model of attempting reimbursements after a data breach or theft.
The platform's cybersecurity infrastructure also identified 350 high-risk addresses linked to investment fraud, reportedly shielding 8,000 users from potential losses. Furthermore, the exchange stated it thwarted over three million credential stuffing attacks throughout the year, a common tactic where hackers use stolen login details from other data breaches.
This development comes as the broader digital asset sector faces relentless threats. The past year saw hackers steal an estimated $3.4 billion, with a growing focus on exploiting vulnerabilities in large crypto entities. Incidents involving ransomware, phishing schemes, and sophisticated malware continue to challenge blockchain security across the ecosystem.
Bybit's system operates by analyzing exchange data for unusual patterns, such as mass withdrawal attempts to new addresses. Withdrawals deemed high-risk are either blocked immediately or trigger a mandatory warning prompt for the user. This allows the operations team to preemptively blacklist dangerous destination addresses, stopping scams before money is lost.
Industry analysts note that implementing AI-based anomaly detection is becoming crucial for defending against hackers who constantly develop new methods to exploit software vulnerabilities, including zero-day flaws. The call for real-time, intelligent threat monitoring is growing louder among cybersecurity professionals.
As the digital finance landscape evolves, the integration of advanced monitoring systems represents a critical layer of defense. Bybit's report underscores the ongoing battle to secure user assets and build trust, demonstrating how proactive measures can directly mitigate financial risk in the volatile world of crypto.
