![\"cara-app-users---jingna-zhang\"](\"https:\/\/www.europesays.com\/ai\/wp-content\/uploads\/2026\/08\/1787512631_870_0x0.jpg\")
<\/p>\n<\/p>\n
User page from Cara, a social network for artists to share their work where AI usage and AI data scraping is explicitly forbidden.<\/p>\n
Courtesy of Cara<\/p>\n
Cara<\/a>, an image sharing social network built specifically for artists who do not consent to have their work used to train generative AI models, suffered a third scraping attack in the past ten days, with copyrighted images and metadata taken from the site without permission appearing on the open source data network Academic Torrents on August 22. This follows previous scrapes of the dataset posted on Reddit (since taken down by the original poster) and the AI model clearinghouse Hugging Face.<\/p>\n \u201cWe have been scraped for a third time. This is now a targeted attack meant to cause artists pain,” wrote Cara founder Jingna Zhang in a thread on X <\/a>Sunday morning. \u201cScrapers know we\u2019re a volunteer project with no financial means to respond. So they say it\u2019s legal, believing themselves to be untouchable.\u201d<\/p>\n \u201cI believe artists who want to share their work should get to do so without violating the principle of consent. I won\u2019t be bullied into making Cara members-only because some people think \u2018consent isn\u2019t always valid.\u2019 You don\u2019t blame victims after harming them.\u201d <\/p>\n She concludes with a plea for community support to raise funds for a legal defense via a GoFundMe<\/a>.<\/p>\n Zhang, a fashion and fine art photographer who was named to the Forbes 30 Under 30 list <\/a>in 2018 and whose work has appeared in Vogue, Elle, Harper\u2019s Bazaar and Time, founded Cara, a public benefit corporation, in late 2022. The network shares some of the same functionality as popular commercial applications like Meta\u2019s Instagram, but has explicit terms of service and features to protect the rights of artists from having their work incorporated into AI training sets without compensation, control or consent.<\/p>\n Artists Seek Refuge from AI<\/p>\n Since it launched, the service has attracted over 1.5 million users, including many professional and hobbyist artists who see AI as a threat to their livelihoods<\/a> and, often, as a basic affront<\/a> to the ideals of art. As working artists, they need to share and promote their artwork to get work from clients and collectors, but they do not want their work used to train systems designed to replace them.<\/p>\n When the first wave of generative image LLMs like Midjourney, Stable Diffusion and Dall-E appeared, they were trained used a huge archive of images, including many under copyright, scraped from the Internet by automated bots. Artists realized that participating in public platforms, even their own personal websites, exposed their work to these systems. Zhang built and launched Cara on a shoestring as a place where artists could share work in an environment where no consent could plausibly be implied. While no site can be perfectly secured in today\u2019s online world, it provides a modicum of legal and technological countermeasures and a clear, human-centric ethical stance.<\/p>\n Jingna Zhang, photographer and founder of the art site Cara<\/p>\n Courtesy of Jingna Zhang<\/p>\n \u201cIn a world where it feel like nobody is actually doing anything, Cara is trying to make a difference,\u201d said Zhang in a phone interview. \u201cIt might not be perfect, it might not solve all the problems or change how the internet is designed, but we are trying, and that makes a huge difference.\u201d<\/p>\n Scrapers defend their actions<\/p>\n Artists\u2019 claims to own and control their own work online are disputed by individuals, groups and commercial entities who believe that advancing the progress of AI entitles them to any and all data they can obtain, regardless of consent. This was apparently the motivation of the original scraper, who posted an archive of nearly 12 million images from Cara to the sub-Reddit r\/DefendingAIArt under the handle \u201cMandarinDrawnPoppy994.\u201d<\/p>\n \u201cScraping is necessary to develop good models. It\u2019s like building a highway \u2013 some houses must be demolished, but in the end everyone benefits,\u201d the poster wrote in a thread titled<\/a> \u201c[AMA] I scraped all of Cara.\u201d<\/p>\n Zhang says she and others reached out to the original scraper and eventually prevailed on him to take down the post. She adds, \u201cnot only did the first scraper delete the dataset, but he has turned around now to offer help, and we\u2019re now co-creating an open source tool <\/a>separate from Cara that will help people check if they’ve been scraped in new datasets in the future.\u201d<\/p>\n Unfortunately, that was not the end of the problem. Several days later, the site was scraped again by a different actor. This time the data was posted on Hugging Face, a hub of resources for AI developers rooted in the open source community, by a poster under the name \u201cIoannis\/Captive Dreamer.\u201d<\/a> <\/p>\n After some people reported the post to Hugging Face Trust and Safety, the team responded<\/a> that \u201cwe have reviewed these [copyright reports] carefully. Because no copies of the artworks are hosted here, and because the URLs [in the dataset] point to the copies the artists published on Cara, there is nothing hosted on Hugging Face that we can disable through our notice and takedown process. This is not a judgement about who owns the works (the artists do); it is about what is stored on our servers. Further copyright reports on the same basis will not change this outcome.\u201d<\/p>\n Hugging Face did not respond to a request for further comment for this story.<\/p>\n Third attack in 10 days<\/p>\n Now on Saturday, August 22, Zhang says the site was scraped for a third time, with the perpetrator taking just 123K images, but also a second data set that includes users\u2019 text posts, bio and information they share on the site. That archive has been posted on Academic Torrents<\/a>, a site that \u201cwas established to meet the demands of science in the age of big data\u201d by providing data for researchers, according to its \u201cAbout\u201d page.<\/a><\/p>\n The series of attacks has shaken the members of the community, despite most users recognizing there was not much that the Cara team could do in the face of escalating capabilities of scrapers and hackers.<\/p>\n