zoom/zoom-e2e-whitepaper

Mention of "Deep Fakes" without definition

Closed this issue · 3 comments

I believe there is a common use of the term Deep Fake technology as mentioned in Phase II, but I think in this context it would be good to define it explicitly.

Could you provide other common uses of Deep Fake Technology? The only definition shown on the first page Google, and available on Wikipedia is:

"Deepfakes are synthetic media in which a person in an existing image or video is replaced with someone else's likeness."

I do not think there will be any misconception on the term Deep Fake technology when it is not explicitly defined in the whitepaper.

mmou commented

Thanks for the note, in our paper we use “deep fakes” to refer to manipulated and/or fabricated audio/video data that uses synthetic media techniques to replace the likeness of one person with another. In the context of video meetings, using a “deep fake” (especially in real-time) of a participant could potentially deceive others about the identity of that participant.

-- Zoom E2E Team

You may want to refer to our paper below which covers exactly this attack, both using "shallow" and "deep" fake technology.

https://sites.uab.edu/saxena/files/2019/12/ss-ccs14-1.pdf

Wiretapping via mimicry: Short voice imitation man-in-the-middle attacks on crypto phones.
Maliheh Shirvanian, Nitesh Saxena.
In ACM Conference on Computer and Communications Security (CCS), November, 2014.