AI Firms Hunt Pre-2022 Books Because AI Slop Wrecked Everything Else, Under NDA
ISBNdb is selling AI firms pre-2022 printed books as slop-free training data, under strict NDAs, because the machines poisoned the open web with their own output.
"The optics problem is real"
404 Media, Emanuel Maiberg
The AI industry flooded the internet with so much generated garbage that its own scrapers now choke on it, so ISBNdb is pitching pre-2022 printed books as the clean fix and calling old paper “structurally guaranteed to be free of this contamination.” The company advertises high-volume acquisition for AI buyers with a legally binding NDA on every engagement, so nobody catches the buyers destroying books during scanning, and its own site admits “the optics problem is real.” A snake that ate the whole library, got sick on its own slop, and now pays a middleman to shred first editions in secret, and that is the f*ucking business model. 404 Media on the pre-2022 book grab
Source: 404 Media, Emanuel Maiberg · Emanuel Maiberg
More from Theft
Patreon Slammed Its Door on AI Scraper Bots and Told Them to Stay Off
Patreon partnered with Cloudflare to block known AI training crawlers at the network level across every post, after its CEO said creators get zero consent, credit, or pay from the models eating their work.
Suno Hack Spilled the Receipt: 113,879 Hours Ripped From YouTube, Called It Fair Use
Hacked source code turned Suno's vague fair-use defense into a line-item inventory: 2,013,545 clips and 113,879 hours from YouTube Music alone, plus Deezer, Genius, and roughly a million hours of podcasts.