Key Insights:
- Bitget Wallet expanded its AI crypto security tools as crypto phishing attacks pushed weekly signing alerts up 7.4x in Q3.
- AI helped identify 90.2% of newly blocked risky domains in Q4 to date, up from 36.3% in Q2.
- Bitget Wallet Chief Operating Officer Alvin Kan said the rise likely reflected both more threats and better detection.
Bitget Wallet expanded its artificial intelligence (AI) security tools on Nov. 11, 2026, after phishing-related signing alerts increased 7.4-fold during the third quarter.
The wallet recorded approximately 37,000 flagged interactions weekly, compared with 5,000 in Q2. However, Chief Operating Officer Alvin Kan said improved detection likely contributed to the increase.
Bitget Wallet Phishing Alerts Jump 7.4x in Q3
Bitget Wallet flagged about 37,000 phishing-related signing requests per week in Q3. The figure stood at 5,000 per week in Q2, as Cointelegraph reported.
Chief Operating Officer Alvin Kan shared the numbers with Cointelegraph. He said the rise likely reflected both more threats and better detection.
The wallet introduced AI-assisted risk detection in Q2, Kan said. Part of the Q3 jump might reflect the newer system. He added that the data could not show how much each factor contributed.
Kan also noted that the tally counted flagged signing interactions. It did not measure unique attacks or confirmed financial losses.
The wallet also split Q3 detections into four risk categories.
Phishing-related signing risks made up 57.4% of Q3 detections. Malicious addresses followed at 16.5%, and risky contracts held 14.2%. Malicious approvals accounted for 11.8%.
Average weekly detections rose across every category from Q2 to Q3. Phishing-related risks grew 7.4x, while malicious addresses grew 6.0x. Risky contracts and malicious approvals rose 5.1x and 4.1x.
How AI Crypto Security Works Inside GetShield
Bitget Wallet ran its checks through GetShield, a real-time security engine. The tool screened websites, permissions, and signing requests before users approved anything.
The AI examined website code and domain registration details for signs of fraud. It also simulated transactions to reveal hidden permissions before a user signed.
GetShield ran three main checks. These covered websites and DApps, permissions and contracts, and signing requests. The last check looked for unexpected changes to a request. Bitget Wallet paired GetShield with external threat intelligence and token approval management.
The AI share of newly blocked risky domains grew each quarter. It stood at 36.3% in Q2 and 70.2% in Q3. In the first four weeks of Q4, it reached 90.2%.
Bitget Wallet also shared how often the AI worked without outside help.
In early Q4, 69.8% of newly blocked domains had no matching external alerts. The share stood at 56.3% in Q3. Weekly blocked domains involving the AI rose 2.5x from Q3 to Q4 to date.
These figures came from Bitget Wallet’s own systems. They were not independently verified, Bitget noted.
The wallet also listed other safeguards. They included encrypted private key storage, multi-party computation and trusted execution environments. Biometric authentication and infrastructure isolation were part of the setup too.
Why Crypto Phishing Attacks Remain a Threat
Kan said, “AI is making cyberattacks cheaper and easier to scale.” He added that self-custodial wallets were moving into payments and everyday finance. The company aimed to spot risks earlier and contain failures. It also aimed to preserve user control.
The Chainalysis 2026 Crypto Crime Report studied AI-linked scams. It found they earned 4.5x more revenue per operation. CertiK’s H1 2026 report showed phishing incidents fell 52.3% year over year. Losses fell only 10.8%.
Crypto phishing attacks also caused real losses this year. In July, a trader lost nearly $1 million to a malicious Ethereum (ETH) token approval. In August, Rapid7 investigated a campaign targeting 885,000 phone numbers.
On Nov. 5, Changpeng Zhao warned users about a fake Trust Wallet website. Researcher Anurag found code on the site that could scan balances and prepare transfers. Signing was disabled when he examined it.
Kan earlier said wallets should show if an approval covered one payment only. He also called for simple ways to revoke permissions.
Bitget Wallet planned to expand AI-driven detection further. It also planned to explore AI-assisted incident response with human oversight.
Users still held the final say on every signature. The wallet’s AI crypto security checks aimed to flag risks before that step.
