He Scraped All of Their Art for AI. Now He’s Collaborating on a Tool to Help Them

Cara, an art platform, has been targeted by scrapers, resulting in three major incidents where large amounts of its publicly available images were stolen. The first incident occurred on August 13, when a user posted a 12-terabyte archive of 12 million works from Cara on Reddit, sparking controversy and debates about the ethics of data harvesting.

Zhang, a key figure from Cara, is currently working on a new open-source tool to protect artists from such attacks. In a surprising turn of events, the individual responsible for the scrapes later agreed to collaborate with Zhang on the new tool. Despite this, other scrapers have continued to take advantage of Cara's vulnerabilities, pulling additional data and uploading it to platforms like Hugging Face and Academic Torrents.

Zhang has launched a GoFundMe to cover legal fees, raising over $100,000 so far, and Cara is actively seeking further legal assistance. Zhang expresses frustration over the scraping and the confusion surrounding what can realistically be done to shield artists from malicious actors. She acknowledges that Cara cannot guarantee complete security and urges users to be aware that no platform is entirely safe from such attacks.

However, Zhang has found an unexpected ally in Heft, the individual who initially scraped Cara, who has since apologized, deleted his dataset, and expressed remorse for the harm caused to the Cara community.