Bitget Wallet Expands AI Checks After Phishing Alerts Rise 7.4x
Bitget Wallet's weekly phishing flags rose 7.4-fold to about 37,000 in Q3. AI now helps identify 90.2% of newly blocked risky domains, though COO Alvin Kan says better detection drives part of the jump.
About 37,000 a week. That is how many signing requests Bitget Wallet flagged for potential phishing in Q3 2026, up from 5,000 in Q2, a 7.4-fold jump in the weekly average. The self-custody wallet is expanding its AI-assisted threat detection in response.
The figures come from COO Alvin Kan and a Nov. 11 company announcement. The expansion covers fraudulent websites, suspicious domains and transaction activity, including requests that can give attackers access to assets without exposing users' private keys.
What the numbers count
Kan was careful about what the tally measures. It counts flagged signing interactions, not unique attacks or confirmed financial losses. The data also comes from Bitget Wallet's own systems.
"The increase likely reflects both greater threat activity and improved detection capabilities, although the data does not allow us to quantify each factor's contribution,"
Alvin Kan, COO, Bitget Wallet
So the 7.4x jump is not a clean read on attack volume. Better detection accounts for some unknown share of it.
AI is doing more of the catching
AI-assisted risk detection was introduced in Q2. Its share of newly blocked risky domains has climbed fast, per the announcement's precise figures:
- Q2: 36.3%
- Q3: 70.2%
- First four weeks of Q4: 90.2%
In early Q4, 69.8% of the domains its AI flagged had no matching alerts in the external threat databases the wallet checked. In Q3 that figure was 56.3%. That is the pitch: the AI is catching sites known threat lists have not yet picked up.
"AI is making cyberattacks cheaper and easier to scale, which means security can no longer rely on known threat lists or a single line of defense,"
Alvin Kan, COO, Bitget Wallet
Where the threats sit
Phishing-related signing requests made up 57% of Q3 detections across four categories. Malicious addresses accounted for 17%, risky contracts 14% and malicious approvals 12%, using rounded announcement figures.
The checks are built into GetShield, the wallet's security engine. They examine website code and domain registration details, and simulate transactions to spot hidden permissions before users sign. Next: wider AI-driven detection and AI-assisted incident response under human oversight, which Bitget Wallet says it plans to explore.
Context
Signing-based theft is the threat these tools target. Recent cases include a trader who lost nearly $1 million in July after signing a malicious Ethereum token approval, and a phishing campaign aimed at 885,000 phone numbers that Rapid7 found in August. Malicious signing requests can drain assets without exposing a private key, so key protection alone does not stop them.
Bitget's wallet unit has been busy on the product side too, as in its recent USD and EUR virtual accounts launch via Noah. The security numbers are self-reported and unaudited. Flagged volume is rising, and the wallet is leaning on AI to keep pace.
