Just so we’re all on the same page with the writer’s strike.
If during the strike, it’s announced about AI generated shows. We are not watching them. Not even out of curiosity. Let them fail every AI generated show they try make.
The human voice can not be replaced by AI. Don’t let them try.
A bit of a TLDR for some questions I saw in the notes:
The team that created Glaze is from the University of Chicago. Their names are each listed in full on the Glaze download website. (This group of students/professors did this for their SPRING BREAK 😱 so go give them some love lol)
It is free to download. No, they won’t ask for or raise money from/for this project.(stated by one of the lead professors of the project).
Glaze is designed to protect artists’ STYLE–which a bunch of ai people have been deliberately fine-tuning their models to mimic (and specifically of current living artists–small or big).
It currently does not protect against composition/trace-like theft (as seen when run through img-to-img) but that would be protected by copyright anyway while STYLE is not.
The University Team has stated that they are dedicated to continuing to improve the tool, like fixing bugs (like overheating older computers by taking up lots of energy when Glazing–it currently runs on CPU so they’re trying to change that to GPU, I believe) and expanding the type of protection given to artists (like working against img-to-img theft).
It currently only works directly on your computer (phones not advised due to current overheating issue, no tablets, or iPads, and no website runthrough since that would be insecure to breaches/scraping/hacks)
It currently works best onpainterly artwork, but can still be used on other forms (team is working on improving this)
IT WORKS BY calculating the changes each image needs for the best protection against style theft by AI, and adds tiny changes throughout the piece, so that your style will, for example, confuse the ai into seeing van gogh. But the ai thieves will see a regular image in your style, feeding it into their model labeled as your work (thus starting the “data poisoning”).
Do not post the original unGlazed piece of your artwork after posting your Glazed version (obviously)
The Team worked directly with over 1,000 artists that were being impacted by the ai theft. Because the team listened to those artists, Glaze accounts for regular art thieves too (i.e. Glaze can’t be removed/cropped etc. like signatures or watermarks when reposted. It’s just part of the image, so even if it ends up on another site and scraped, the Glazing is still in effect)
When you run your artwork through Glaze, no information is sent back to the Team. (Aka, no scraping on their part. The app receives information from the Team (like updates) but no information from you is given to them through the app. Basically Team servers —> You and NOT Team servers <-–>You) One-way data street.
Brief misunderstanding happened over an open-source license for the front-end part of the app. (Used open-source coding for front-end, not knowing that code’s use-license states it is only for other open-source uses, not closed-source (the back-end code of the app is private to prevent counter-counter measure developments)). The Team took down the app until they replacedthe front-endcode with code written from scratch by the team. They are now not in violation of that open-source license since they are no longer using it. (you have 30 days to remedy a license breach once informed; they did so in 2)
The Team is currently in touch with Japanese artists to better expand the tool for use to protect their art styles
From what I understand of it, Glaze is an AI tool designed to be anti-AI (Think Arnold Schwarzenegger in Terminator 2: one Terminator robot vs. all the other Terminators 😂)
You can download it from their website and also contact them through email there with any questions, problems, or bugs. The website: https://glaze.cs.uchicago.edu/
reblogging this every fucking time it comes across my dash
I feel like I’m going to go to my grave without figuring out if Dr. Horrible is deliberately a condemnation of the geek-flavored version of toxic masculinity that would, years later, play a significant part in the resurgence of open white nationalism and the like in America, or if Joss Whedon is just a dumbass who wrote an extended callout post for himself on accident
Personally, I think it being an accidental self-callout enriches the narrative immensely. There’s layers to writing a cliche villain who thinks he’s the hero (but actually ends up being the villain) (but the writer thinks he’s a hero) (but the writer is wrong, possessed by the same self-delusions the villain has). It’s a text that doesn’t know what it’s about in a way that mirrors the way the main character doesn’t know what he really is. It fascinates me. At this point I’d actually be disappointed if Joss did it on purpose.
the reason none of us can ever leave tumblr is because we’ve already evolved to having this be our only suitable habitat. we’re the devils hole pupfish of people.
my homunculus just wasted like $60 in ink printing out all these fucking soyjack edits of me. how about ‘that feel when i dont feed you for a week’ you little asshole
If you are taking the car to Chicago, can you even call yourself a feminist. There are Women on that train who want to have gay sex with you but can’t because you are taking a fucking car to Chicago
Right now there is a very buff Butch woman all alone in the Cafe car and she could be having steamy gay sex with you and be moving into your apartment if you were just willing to take the train to Chicago
I’m busy trying to get more lesbians to ride the train instead of taking their Subarus to Chicago
my favorite dril thing is when he’ll just RT something from an account that hasn’t posted in six years and it’s clear he just came up with a stupid @ and checked to see if anyone had it
ALT
