OpenAI, Google, Microsoft and Anthropic announce shared AI training preference signal
The robots.txt token would let websites express a generative AI training preference without blocking crawlers.
TLDR
The companies announced on October 30, 2026, that they would honor __usage_ai-training in robots.txt, allowing sites to express a preference against using their content for generative AI training. The signal does not block crawling or remove material from existing models. Anthropic has begun implementing support; OpenAI aims to follow in the coming months, Google in 2027 and Microsoft starting February 1, 2027.
Combined views
—
7 Sources, first seen ago


