From early 2023 A small group of volunteers and photographer Jingna Zhang have tirelessly worked to maintain an online image-sharing site. social media App called Portfolio Cara. It has so far attracted around 1.5 million musicians. Why did they join the platform? The shared opposition? unauthorized use of their work They want to use AI to create models of their own and they don’t want to be exploited.
But while Cara filters out AI images and offers protective features—including Glaze, a tool meant to mask the style of the images picked up by scrapers in order to disrupt AI mimicry—preventing scrapes themselves is nearly impossible. Cara has been subjected, starting on August 13th, to 3 major scrapes. This has increased its server costs and worried creators that had moved from Instagram to Cara. Instagram makes all of the content available for Meta to use as training data.
The first of these incidents came to light when the individual responsible posted a 12-terabyte archive of 12 million works from Cara—more or less its entire library of publicly available images—on the subreddit r/DefendingAIArt. “It was a fun project,” wrote the redditor, MandarinDawnPoppy994, in his since-deleted post, saying the process cost him less than $10.
“We actually found out about it through our users tagging us,” Zhang tells WIRED the scraper is no longer in use. “gloating and looking for other people to join him to do something with the dataset on Reddit,” He sparked an intense debate on AI-related forums regarding the ethics behind his work. “I just feel it’s targeted and very hurtful,” “She adds,” she notes. “laws are not caught up on” Because scrapers are often able to justify their activities as being technically legal, they can do so. Zhang is also a part of separate class actions filed by artists against Stability AI and Midjourney and Google. The visual artist’s claim that these companies were using their image-generating tools to train them on copyrighted works.
The person who stole all of Cara’s art would regret the stunt, and will now collaborate with Zhang to create a tool that protects artists.
Unfortunately, in the interim, scrapers took advantage of Cara’s vulnerability and limited resources. AI supporters have tended to be a bit sceptical about the technology. objected MandarinDawnPoppy994 encouraged a select few to take on Cara in order to achieve what Zhang views as right. “copycat” attacks.
The second scraper is pulled out 8.5 million links Cara can also be used to store metadata, such as user names, titles and tags. uploaded Hugging Face is the AI platform for developers. Hugging Face responded to the takedown requests after being bombarded by them. statement While it would send a notification to the user “CaptiveDreamer,” It could not remove URLs and personal metadata at the same time. “no copies of the artworks are hosted here,” Links “point to the copies the artists published on Cara.” This is what the company said “further copyright reports on the same basis will not change this outcome.”
The third scraper, who obtained the images and text from Cara on August 22nd, also shared user bios, which included some personal information. Academic Torrents. Zhang launched an GoFundMe For legal fees she set a target of $120,000 and explained that it would be used to explore all possible strategies for defending Cara using copyright and cyber laws. Cara has been actively seeking additional legal support.

