Bitget Wallet expanded its AI-powered threat detection after average weekly signing requests flagged for potential phishing rose 7.4-fold from 5,000 in Q2 to about 37,000 in Q3 2026. The tally covered flagged signing interactions rather than unique attacks or confirmed financial losses, and Chief Operating Officer Alvin Kan said the increase likely reflected both greater threat activity and improved detection, without quantifying either factor's contribution. Phishing-related signing risks represented 57% of detections across four monitored categories in Q3, followed by malicious addresses at 17%, risky contracts at 14% and malicious approvals at 12%. The company said its proprietary AI identified 90% of newly blocked risk domains in October, compared with 36% in Q2; the existing record gives more precise figures of 90.2% in the first four weeks of Q4, 36.3% in Q2 and 70.2% in Q3. Nearly 70% of the October domains, or 69.8% in the earlier detailed figures, had no corresponding alerts in external threat-intelligence databases. Integrated into the GetShield security engine, the system analyzes website code, domain-registration data and transaction outcomes to identify fraudulent sites, emerging threats and hidden permissions before users authorize transactions. Bitget Wallet also cites encrypted private-key storage, multi-party computation, trusted execution environments, biometric authentication, infrastructure isolation, third-party assessments and continuous vulnerability monitoring as layers in its security framework. The self-custodial wallet plans to expand AI monitoring and explore AI-assisted incident response under human oversight.