AI Crypto Scams Jump 1,400% in 2025

• Impersonation scams using AI technology rose by 1,400% in 2025, leading to a record-breaking loss of $17 billion in crypto fraud.
• Bybit intercepted or recovered $300M of $500M in suspicious withdrawal requests in Q4 2025, safeguarding over 4,000 users from losses.
• Bybit’s three-tier Dynamic Risk-Based Protection System prevents fraud from occurring before it reaches the next platform.

The number of AI crypto scams has risen 1,400% in 2025, causing global fraud losses to reach a record high of $17 billion, Chainalysis reported. 

The recent rise in AI-powered impersonation attacks has led exchanges to completely forgo reactive security strategies. Bybit is one of the exchanges that are at the forefront of this fight. 

In the last quarter of 2025, the exchange was able to intercept or recover $300 million out of $500 million in suspicious withdrawal attempts flagged as potential scams, safeguarding over 4,000 users.

AI-Enabled Impersonation Scams Drive Record Fraud Numbers

The AI crypto scams have brought a paradigm shift in the manner in which the frauds function within the market in 2025. The scammers are utilizing artificial intelligence to imitate the genuine platforms with a great degree of speed and accuracy. 

The impersonation attacks are difficult to identify and more sophisticated than the phishing attacks. According to Chainalysis, a total of $17 billion in crypto was lost to scams and fraud in 2025. This is a record high, mainly due to the adoption of AI tools by the scammers.

 The 1,400% surge in AI-enabled impersonation cases alone tells the story of how fast the threat has escalated. As CryptosRus noted on X, “AI scams are exploding—but defenses are getting smarter.” The “race between the scammers and the exchanges” has emerged as one of the most prominent security struggles of the year in the crypto world. 

Exchanges that fail to keep up with the times are increasingly putting their users at risk of sophisticated and rapidly evolving threats.

Exchanges Shift From Reactive to Proactive Security Models

The traditional approach of tracing funds after losses have occurred is no longer sufficient against AI-powered fraud. Bybit responded to this reality by building a Dynamic Risk-Based Protection System structured around early interception. 

The framework operates through three tiers, each addressing threats at different levels of risk intensity. The first tier flags low-risk anomalies, such as withdrawal clusters sent to newly created wallet addresses. 

Automated behavioral checks and big-data heuristics are deployed at this stage to catch emerging patterns. High-risk destinations can also be preemptively blacklisted before any transaction is processed.

The second tier delivers real-time warnings at the point of withdrawal when an account is linked to suspicious activity. 

This interruption is specifically designed to break the psychological urgency that social engineering scams exploit. Stopping a victim mid-transaction is often the most effective window to prevent an irreversible loss.

Cooling-Off Period Targets AI-Driven Pig Butchering Schemes

The third tier immediately blocks withdrawals connected to confirmed scam addresses in real time. A mandatory one-hour cooling-off period follows, giving users a window to reconsider before funds move permanently. 

This protocol has become one of Bybit’s most effective tools against long-form investment fraud in 2025. Pig butchering schemes rely on weeks of psychological conditioning before requesting any transfer of funds. 

The AI tools have increased the speed of these scams and made them harder to identify. The cooling-off period interferes with the last and most important step of this type of fraud. David Zong, Bybit’s Head of Group Risk Control, outlined the platform’s broader security philosophy clearly. 

He stated, “Our mission is to transform risk control from a silent shield into an active, intelligent guardian.” That shift reflects a wider industry movement toward behavioral-based, upstream security across major exchanges.

Cross-Chain Monitoring and Shared Intelligence Strengthen Industry Defenses

Bybit’s Q4 2025 results reveal how deep the fraud problem runs across the crypto ecosystem. The exchange identified 350 high-risk fraud addresses through AI-driven on-chain monitoring during the quarter. Additionally, over three million credential stuffing attacks were blocked within that same period.

Beyond its internal systems, Bybit integrates real-time intelligence from TRM Labs, Elliptic, and Chainalysis. These partnerships create a standardized risk identification layer applied to both deposits and withdrawals. 

Cross-chain tracing across bridges and mixers further extends coverage as illicit funds fragment across networks.

Fraud mapping and address tagging are becoming shared efforts rather than competitive secrets among major platforms. 

CryptosRus reiterated this on X, urging users to check URLs, turn on two-factor authentication, and not respond to unsolicited support messages. 

Despite record interceptions, the $300 million Bybit recovered is only a fraction of the $17 billion lost worldwide in 2025.