Re:Selfie-based authentication raises eyebrows among infosec...
by patrakov on Saturday July 13, 2024 @04:20AM
(#40952617)
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.
[ Parent
| on HN ]
Re:Selfie-based authentication raises eyebrows among infosec...
by idiotsecant on Saturday July 13, 2024 @07:59AM
(#40953554)
I 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.
Re:Selfie-based authentication raises eyebrows among infosec...
by vasco on Saturday July 13, 2024 @04:29AM
(#40952648)
Nothing a little faceswap connected to the webcam feed can't fix.
Re:Selfie-based authentication raises eyebrows among infosec...
by lazide on Saturday July 13, 2024 @08:02AM
(#40953569)
Yup, but it’s an arms race.
They will come next.
Re:Selfie-based authentication raises eyebrows among infosec...
by vasco on Saturday July 13, 2024 @08:39AM
(#40953747)
It'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?
Re:Selfie-based authentication raises eyebrows among infosec...
by lazide on Saturday July 13, 2024 @09:47AM
(#40954103)
At 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.