Glossary
AI training opt-out
A machine-readable signal — robots.txt rules, ai.txt files, or embedded metadata — asking crawlers not to use a site’s content for AI training. An opt-out limits scraping; it is not a licensing mechanism in itself.
Working mechanisms include robots.txt rules for AI crawlers such as GPTBot and Google-Extended, noai metadata tags, and licence-reservation statements. Under the EU’s Article 4 regime a machine-readable reservation has legal force; elsewhere it is a request that reputable crawlers honour. Opt-outs stop future scraping but do not un-train existing models, and they generate no revenue by themselves.
Why it matters
An opt-out preserves the licensing option; a licence is what exercises it.