Google has announced it is introducing face verification video as a new sign-in method for Google accounts.
While still able to access the account normally, a user records and registers a short video of themselves using a phone or computer. Later, when locked out, they record a new video that is compared against the registered one to confirm identity.
The design functions primarily as a recovery mechanism — for forgotten passwords or lost two-factor devices.
The core mechanism is liveness verification. Actions such as turning the head and nodding confirm the footage is being captured in real time, blocking deepfake impersonation.
The structure verifies whether a real person is in front of the camera right now, so a single photograph or a pre-generated deepfake video cannot pass.
In an environment where generative AI has collapsed the cost of producing convincing face video, the design reflects a judgment that static image comparison no longer constitutes authentication.
Google said registered videos are encrypted and stored for sign-in support purposes, and that users can delete them at any time.
The company also acknowledged the privacy burdens:
The first is the fundamental problem of biometrics generally. A leaked password can be changed; a face cannot. The second will be encountered far more often in practice — failing on lighting conditions precisely when you need account recovery leaves no fallback.
Account recovery experience affects churn. A failed sign-in is an exit. Google expanding recovery options reflects a recognition that authentication friction is a business problem.
Design the fallback alongside the biometric. If lighting or appearance changes can cause failure, what happens on failure is half the design. Depending on a single authentication method is the risk.
Account security requirements are spreading into ad operations too. For the same trend in advertising platform access, see Google Makes Passkeys Mandatory for the Google Ads API.
Account for consumer sentiment around biometric collection. Google leading with encryption and deletion rights suggests it judges trust, not technology, to be the main barrier to adoption.
AI destabilising identity verification itself carries brand risk as well — discussed in How Prompt Injection Threatens Your Brand.
Users register a short video of themselves while able to access their account normally. If later locked out, they record a new video that is compared against the registered one to verify identity.
Through liveness verification — actions such as turning the head and nodding confirm the footage is being captured live, so a photo or pre-made deepfake video cannot pass.
Google says it is encrypted and retained for sign-in support purposes, and that users can delete it at any time.
Facial information cannot be reset like a password, and verification may fail due to lighting, camera quality or changes in appearance.
To apply what you just read to your own site, start with a free audit of where things are now.
A strategist replies within 24 hours on business days.
Opening an AI before a search engine is becoming the default. As the path back to original creators narrows, where does competitive advantage in the AI era actually come from?
Meta expanded its vibe coding app Pocket to all U.S. users. Describe what you want and it builds a small app or game — then that output circulates in a social feed.
Meta's CEO published a 6,500-word essay on personal superintelligence that doubles as an argument against AI regulation — while Meta cut 20% of staff earlier this year.