Google has rolled out a new way to get back into a locked account, and it doesn’t involve a password reset link or a security question about your childhood pet. The feature, called selfie video, launched on July 23 and is now rolling out gradually to eligible personal Google Accounts worldwide.
The mechanics are straightforward. A user sets up the feature in advance by recording a short video of themselves through their phone’s front camera or a computer webcam, following on-screen prompts to turn their head in different directions. That enrollment video becomes the reference point. If the person later loses access to their account, whether through a forgotten password, a lost device, or a sign-in attempt from an unrecognized device, Google asks for a fresh selfie video and compares the two to confirm it’s the same person.
The timing isn’t incidental. Synthetic media has gotten good enough, and cheap enough to produce, that traditional identity checks are increasingly vulnerable to being spoofed by AI-generated images and videos.
Google says the system layers multiple security checks, including AI-powered facial matching, liveness detection, guided movements, and suspicious sign-in detection, specifically to reduce the risk of impersonation using photographs, replayed videos, or deepfakes.
The guided head-turning requirement during both setup and recovery exists precisely to defend against static photos and AI-generated fakes that can’t replicate live, directed movement in real time.
Privacy details matter here, and Google has been fairly specific about them. Selfie videos are encrypted while stored, and by default they serve exactly one purpose: signing in. Whether Google can use a person’s video to improve its broader facial recognition and verification technology is an opt-in choice, not something enabled automatically. Users can delete their video at any time through Google Account settings. However, Google notes it may retain deleted footage temporarily for security purposes, and longer in cases tied to policy enforcement.
The rollout does have limits. The feature currently excludes Google Workspace accounts, child accounts, and anyone enrolled in Google’s Advanced Protection Program, the company’s highest-security tier typically used by journalists, activists, and public figures who face elevated risk of targeted attacks. It also isn’t replacing anything. Selfie video sits alongside recovery phone numbers, recovery emails, passkeys, and recovery contacts rather than replacing any existing method, giving users one more door back into their account rather than a single new gatekeeper.
There’s a regulatory backdrop worth noting too. European users will encounter this feature against a shifting legal landscape, since new biometric categorization and deepfake disclosure obligations take effect there on August 2, 2026, meaning Google’s rollout in the EU will need to operate inside rules specifically designed to govern exactly this kind of technology.
Whether biometric verification actually stays ahead of deepfake sophistication long-term is a fair question, and one that security researchers are already raising. Face-matching systems have been fooled before, and as generative video tools improve, the arms race between verification and impersonation is unlikely to have a permanent winner. What Google is betting on, for now, is that liveness detection and layered checks make forgery significantly harder than it is with a static photo or a pre-recorded clip, even if “harder” doesn’t mean “impossible.”
The feature builds on a broader security push from Google this year. Last month, the company introduced fake call detection on Android.
This tool flags suspected impersonation calls in real time as voice-cloning scams have become more sophisticated and harder to distinguish from genuine calls. Selfie video extends that same logic, defending against synthetic identity, to the account recovery process itself.
