Re:Image of a genome autosuspends you from Twitter
by tsaprailis on Monday May 30, 2022 @04:45PM
(#31563451)
Oce you upload the image, you get immediately suspended with the justification that the post contains personal information. The weird thing is the image contains no text whatsoever, or anything for that matter that could associate the image as a genome.
This is the COVID genome btw.
Re:Image of a genome autosuspends you from Twitter
by webmaven on Monday May 30, 2022 @07:15PM
(#31564623)
Hmm. The "personal information" bit doesn't fit particularly well .
I'd bet on the image running afoul of something like an anti-steganography filter (except in this image the dots aren't actually hidden from human perception), and the "personal info" line is either an error or is intended as misdirection.
I mean, I suppose the unsavory folks doxxing others may be passing around and broadcasting that personal info in steganographic images (or, you know, as ordinary images), but I don't think that makes a lot of operational sense.
Re:Image of a genome autosuspends you from Twitter
by robswc on Monday May 30, 2022 @05:44PM
(#31563964)
Very interesting...
Not at all qualified to even speculate on this so I would love to see what someone might think causes it. Surly a bug with a bot right? (vs someone manually setting a flag)
Re:Image of a genome autosuspends you from Twitter
by devwastaken on Monday May 30, 2022 @10:31PM
(#31565570)
With Discord I could not send an image of a car because it was considered "adult". Automatic systems are not at all up to the task of properly identifying new images, only known existing hashes.
Letters maybe, but its not standard to represents genomes as dots. Thats just an "artistic" representation.
I'd bet on the image running afoul of something like an anti-steganography filter (except in this image the dots aren't actually hidden from human perception), and the "personal info" line is either an error or is intended as misdirection.
I mean, I suppose the unsavory folks doxxing others may be passing around and broadcasting that personal info in steganographic images (or, you know, as ordinary images), but I don't think that makes a lot of operational sense.
Not at all qualified to even speculate on this so I would love to see what someone might think causes it. Surly a bug with a bot right? (vs someone manually setting a flag)