You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi Whonix dev here. Whonix is a privacy distro based on Tor similar to Tails but leverages VMs to provide app isolation. Our goal is to assemble a toolbox to protect activists' anonymity and we realize that masking location IP while an important step, is not enough to protect our users from advanced adversaries that use stylometry.
We are in the process of including a tool called Kloak (by researcher Vinnie Monaco) for masking keystroke timing and so a program like yours would be another essential piece of the puzzle to foil deanonymization based on semantic analysis.
Unfortunately the widely publicized anti-stylometry tool "Anonymouth" has many serious bugs and its repo has been dead for years. We would greatly appreciate if you work with us on getting your software to run on Debian so we can distribute it to our users.
If you are interested please see orjail/orjail#27 for an example of how orjail got its Debian packaging. Also our lead dev Patrick would welcome giving input on the process.
Questions:
How does your model fare against an anonymity adversary using the Writeprints-Static Approach mentioned in Greenstadt's [0] paper?
In that same paper machine translation was criticized for its low success rate in stumping classifiers while also giving garbled output with bad grammar. Do you use the term "machine translation" in the same way or is it just a different approach that is automated?
Hi Whonix dev here. Whonix is a privacy distro based on Tor similar to Tails but leverages VMs to provide app isolation. Our goal is to assemble a toolbox to protect activists' anonymity and we realize that masking location IP while an important step, is not enough to protect our users from advanced adversaries that use stylometry.
We are in the process of including a tool called Kloak (by researcher Vinnie Monaco) for masking keystroke timing and so a program like yours would be another essential piece of the puzzle to foil deanonymization based on semantic analysis.
Unfortunately the widely publicized anti-stylometry tool "Anonymouth" has many serious bugs and its repo has been dead for years. We would greatly appreciate if you work with us on getting your software to run on Debian so we can distribute it to our users.
If you are interested please see orjail/orjail#27 for an example of how orjail got its Debian packaging. Also our lead dev Patrick would welcome giving input on the process.
Questions:
How does your model fare against an anonymity adversary using the Writeprints-Static Approach mentioned in Greenstadt's [0] paper?
In that same paper machine translation was criticized for its low success rate in stumping classifiers while also giving garbled output with bad grammar. Do you use the term "machine translation" in the same way or is it just a different approach that is automated?
References:
[0] https://web.archive.org/web/20130806053421/https://www.cs.drexel.edu/~sa499/papers/adversarial_stylometry.pdf
cc/ @adrelanos
The text was updated successfully, but these errors were encountered: