Google, OpenAI, and Anthropic back a shared robots.txt AI training signal
Google, OpenAI, Microsoft, and Anthropic announced on October 30, 2026 that they will support a single shared robots.txt token, __usage_ai-training, letting site owners set one AI training preference instead of a separate token for each provider. It also gives Bing an opt-out it did not have before, since sites previously had to allow Bing to train on their content to appear in search results. The signal is voluntary and not enforceable, and rollout is staggered — only Anthropic honors it today, OpenAI plans to follow in the coming months, Microsoft in February 2027, and Google in early 2027 — so sites must keep using the old per-provider tokens for now.
SOURCES
- Hacker News (AI)2026-11-06 17:56