Updated
Updated · WIRED · Aug 28
Jingna Zhang, Ex-Scraper Build Lantern After 12 Million Cara Images Were Harvested
Updated
Updated · WIRED · Aug 28

Jingna Zhang, Ex-Scraper Build Lantern After 12 Million Cara Images Were Harvested

2 articles · Updated · WIRED · Aug 28

Summary

  • Lantern is an open-source tool Zhang and former scraper Heft are building to fingerprint artists’ images and alert them when their work appears in public AI datasets.
  • Heft joined the effort after scraping roughly 12 million public Cara images into a 12-terabyte archive, then apologizing when artists described panic attacks and deleted portfolios.
  • Cara, which has about 1.5 million artists, was hit by three major scrapes from Aug. 13 to Aug. 22, including one leak of 8.5 million links and another of 123,000 images plus personal data.
  • Those attacks drove up server costs, pushed some users off the platform, and led Zhang to seek $120,000 for legal fees; more than $100,000 had been raised by Thursday.
  • Zhang and Heft say no site can be made fully unscrapable, casting Lantern as a post-scrape workaround while artists wait for stronger legal and regulatory protections.

Insights

A hacker exposed millions of files before switching sides; can their new open-source fingerprinting tool actually track stolen art across the web?
If no public website is truly unscrapable, how can creators survive the relentless AI data harvest without vanishing from the internet entirely?