Keplr Wallet Losses Put Social-Login Recovery Trust in Focus

Originally reported by Liu Jiaolian on November 10, 2026, at 14:04 (UTC+8). The author reported that assets had been taken from several thousand Keplr social-login wallet accounts. According to the article, Keplr attributed the problem to a vulnerability in Web3Auth’s infrastructure. The technical cause had not been disclosed in detail, and the author noted disagreement between the providers. The account-loss figure and attribution are reported claims, not a confirmed explanation of how the wallets were compromised.

The article compares four ways to safeguard or recover wallet keys. Its baseline is a seed phrase kept by the user on paper: no recovery provider holds it, but loss, theft or damage can leave the user without access. For Uniswap Wallet, the author describes an encrypted seed-phrase backup in the user’s iCloud or Google Drive. Recovery requires access to the cloud account and a separate encryption password, adding dependence on the account and cloud service without giving the provider the plaintext phrase under the described design.

In the author’s account of MetaMask social login, an encrypted seed phrase is held on MetaMask servers, while the decryption key is split among server nodes. Recovery requires a Google, Apple or Telegram login and a user-set password. The author argues that if attackers obtained the key shares, they could try passwords offline, outside the rate limits of normal online recovery. That is a risk scenario, not a reported breach of MetaMask.

For the Keplr arrangement discussed in the article, the author describes three key shares—one on the user’s device, one on a cloud server and one for recovery—with any two sufficient to reconstruct the key. The author speculates that two shares effectively controlled by one provider could create a single point of compromise, but explicitly says the Keplr incident’s actual mechanism awaits further technical disclosure.

These are key-recovery arrangements, not the same as an exchange taking full custody of customer assets. The question for each is which parties can participate in recovery—and whether the user can verify that the stated separation of control holds in practice. The reported Keplr losses alone do not establish which trust boundary failed.

Source: techflowpost.com (originally published 2026-11-10 14:04 UTC+8)