Bitget Wallet Expands AI-Powered Threat Detection to Strengthen Wallet Security
Announcement
AI is changing the nature of crypto security. Attackers are using increasingly sophisticated tools to create convincing phishing websites and deceptive transactions, while traditional threat databases cannot always identify risks that have yet to be reported.
To address this gap, Bitget Wallet has expanded its AI-powered threat detection capabilities, combining AI-assisted website and domain analysis with GetShield's pre-sign security checks. These capabilities help identify emerging phishing risks and risky signing requests before you authorize potentially harmful transactions.
During the first four weeks of Q4 2026, our proprietary AI contributed to identifying
90.2% of newly blocked risk domains, which exemplifies how AI is becoming a larger part of domain-risk monitoring.
Identifying emerging threats beyond known databases
The Q4 figure continues a quarter-by-quarter rise in AI involvement. Our proprietary AI participated in identifying
36.3% of newly blocked risk domains in Q2,
70.2% in Q3, and
90.2% in Q4 to date.
AI also expanded detection coverage beyond the two external threat intelligence databases used for comparison. In Q4 to date,
69.8% of newly blocked risk domains were flagged by our proprietary AI without corresponding risk flags from those databases, up from
56.3% in Q3.
Phishing-related signing risks accounted for
57.4% of Q3 detections across four monitored categories, followed by malicious addresses at
16.5%, risky contracts at
14.2%, and malicious approvals at
11.8%.
From Q2 to Q3, average weekly detections increased
7.4 times for phishing-related risks,
6.0 times for malicious addresses,
5.1 times for risky contracts, and
4.1 times for malicious approvals.
AI-supported domain blocking also expanded: the weekly average of newly blocked risk domains involving our proprietary AI increased
2.5 times from Q3 to Q4 to date, or
5.7 times compared with Q2.
Industry data provides broader context. The
Chainalysis 2026 Crypto Crime Report found that scams with identifiable onchain links to AI vendors generated
4.5 times more revenue per operation than scams without those links. Meanwhile,
CertiK's H1 2026 report found that phishing incidents declined
52.3% year over year, while associated losses fell by only
10.8%.
How AI-powered threat detection works
Our AI-assisted analysis examines website code patterns and suspicious domain characteristics alongside traditional threat intelligence databases to identify emerging phishing risks.
GetShield, Bitget Wallet's proprietary real-time security engine, combines AI-assisted website analysis with three layers of pre-sign checks for supported wallet interactions:
-
Website and DApp screening: checks websites and decentralized applications for potential risks.
-
Permission and contract analysis: analyzes signing permissions and smart contracts before authorization.
-
Signing request checks: detects unexpected changes to signing requests.
Together with token approval management and external threat intelligence, these capabilities help you identify suspicious interactions and potentially dangerous permissions before signing while retaining control of your assets.
Protection across every layer
Security can no longer rely on known threat lists or a single line of defense. Beyond threat detection, our security framework includes encrypted private-key storage, multi-party computation, trusted execution environments, biometric authentication, infrastructure isolation, third-party risk assessments, and continuous vulnerability monitoring. These measures address different security risks across wallet operations and are designed to limit the potential impact of individual failures.
We plan to further expand AI-driven threat detection and explore AI-assisted incident response to identify emerging risks and accelerate responses under human oversight, with a focus on identifying risks earlier, strengthening safeguards at different layers.
Visit the
Bitget Wallet security page to learn more.
Known threat lists can't flag what they haven't seen. In October, our AI helped identify 90% of newly blocked risk domains. Nearly 70% had no external threat intelligence alerts. We're expanding AI-powered threat detection with GetShield. Learn more:
https://web3.bitget.com/en/about/security-technology
2026-11-10
Recommended
