4 ms·
Speaking as GCash user: sometimes the instructions say "blink", sometimes "stay still", sometimes "look right", and it is thus not possible to predict the timin
by patrakov 2y ago
Speaking as GCash user: sometimes the instructions say "blink", sometimes "stay still", sometimes "look right", and it is thus not possible to predict the timing to a sufficient degree to pass it with a pre-recorded video.
- vasco 2y agoNothing a little faceswap connected to the webcam feed can't fix.
- lazide 2y agoYup, but it’s an arms race. They will come next.
- vasco 2y agoIt's funny to extend this to the max. Prick my finger to take a blood sample into a DNA sequencer enclave within the processor of each device?
- lazide 2y agoAt some point I expect a physical presence at a given location is what will matter. If someone identifiable has to physically be at, say, a police station, or DMV, it cuts out 95% of the complexity involved in all the security stuff we’re dealing with. Just from an attack scalability perspective.
- idiotsecant 2y agoI guess the solution is to render fake video on the fly doing pose and expression emulation which is possible with a few static shots, open source tools, and a halfway decent machine. It might not fool a human, but you dontnhave too and I suspect it won't be long before fooling the average human will be pretty easy too.