AI labs' distillation complaints reopen the web-scraping fair-use fight

Business Insider says the latest dispute over model distillation exposes an awkward symmetry for AI labs. Anthropic, OpenAI, and Google have warned that rivals can use model outputs to reproduce expensive capabilities, while publishers and site owners have long argued that AI companies extract value from public web content without permission. The piece also points to Cloudflare-style crawl-versus-referral concerns and the unsettled legal line between benign distillation and extraction attacks.

Key details: Anthropic has complained about rivals using its model outputs for distillation, OpenAI and Google have made similar warnings about output extraction, Publishers and site owners have raised parallel complaints about AI crawlers and web scraping.

Why it matters: The same legal and economic logic behind AI training disputes is now turning inward as labs try to protect their own model outputs.